github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-21 13:11:39 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #12781] After updating from ollama 0.9.3 gpu offloading stopped working on models larger than VRAM #8478

New Issue
Closed
opened 2026-04-12 21:10:09 -05:00 by GiteaMirror · 37 comments
GiteaMirror commented 2026-04-12 21:10:09 -05:00
Owner
Copy Link

Originally created by @kiliansinger on GitHub (Oct 25, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12781

What is the issue?

Working on windows 11, with RTX 4070 8GB VRAM I successfully used 13GB models such as Qwen3-Coder-30B-A3B-Instruct-1M-Unsloth:UD-IQ3_XXS. After upgrading to newer versions of ollama this feature stopped to work, rendering the use of these models too slow on typical CPUs. This issue is related to the closed issue 11676. There it was proposed that the issue is resolved by upgrading from older cuda v11. But I was already on cuda v13.

Proposed PR https://github.com/ollama/ollama/pull/12856

Relevant log output

OS

Windows

GPU

Nvidia

CPU

AMD

Ollama version

0.9.3

Originally created by @kiliansinger on GitHub (Oct 25, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12781 ### What is the issue? Working on windows 11, with RTX 4070 8GB VRAM I successfully used 13GB models such as Qwen3-Coder-30B-A3B-Instruct-1M-Unsloth:UD-IQ3_XXS. After upgrading to newer versions of ollama this feature stopped to work, rendering the use of these models too slow on typical CPUs. This issue is related to the closed [issue 11676](https://github.com/ollama/ollama/issues/11676). There it was proposed that the issue is resolved by upgrading from older cuda v11. But I was already on cuda v13. Proposed PR https://github.com/ollama/ollama/pull/12856 ### Relevant log output ```shell ``` ### OS Windows ### GPU Nvidia ### CPU AMD ### Ollama version >0.9.3
GiteaMirror added the needs more infobug labels 2026-04-12 21:10:09 -05:00
GiteaMirror commented 2026-04-12 21:10:11 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

Server log will help in debugging.

<!-- gh-comment-id:3447891209 --> @rick-github commented on GitHub (Oct 26, 2025): [Server log](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) will help in debugging.
GiteaMirror commented 2026-04-12 21:10:11 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

Here is the server log starting

ollama run danielsheep/Qwen3-Coder-30B-A3B-Instruct-1M-Unsloth:UD-IQ3_XXS

in the following presented for ollama 0.12.6 (failing approx. 95% of cases), then for ollama 0.9.3 (running stable 100% of cases).

Interestingly I also found that in 5% of cases (after rebooting) ollama 0.12.6 would run through. This is also listed as third log.

Here are the log outputs

  • with ollama 0.12.6 failing:
time=2025-10-26T17:52:03.657+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-26T17:52:03.677+01:00 level=INFO source=images.go:522 msg="total blobs: 32"
time=2025-10-26T17:52:03.679+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0"
time=2025-10-26T17:52:03.680+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)"
time=2025-10-26T17:52:03.680+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..."
time=2025-10-26T17:52:05.280+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v13 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB"
time=2025-10-26T17:52:05.281+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB"
[GIN] 2025/10/26 - 17:52:05 | 200 |         562µs |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 17:52:05 | 200 |      2.2054ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/26 - 17:52:36 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 17:52:36 | 500 |            0s |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 17:52:45 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 17:52:45 | 200 |     32.3612ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T17:52:46.148+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T17:52:46.148+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T17:52:46.148+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T17:52:46.148+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T17:52:46.149+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50809"
time=2025-10-26T17:52:46.152+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T17:52:46.152+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T17:52:46.152+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T17:52:46.152+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T17:52:46.152+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.3 GiB"
time=2025-10-26T17:52:46.152+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T17:52:46.194+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T17:52:46.211+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50809"
time=2025-10-26T17:52:46.220+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T17:52:46.236+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T17:52:46.320+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T17:52:46.548+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T17:52:46.570+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T17:52:46.969+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T17:52:46.970+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50809/load\": read tcp 127.0.0.1:50816->127.0.0.1:50809: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 17:52:46 | 500 |    1.1105028s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T17:52:46.995+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
  • with ollama 0.9.3 running through 100% reliable
time=2025-10-26T09:40:18.531+01:00 level=INFO source=routes.go:1235 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-26T09:40:18.553+01:00 level=INFO source=images.go:476 msg="total blobs: 32"
time=2025-10-26T09:40:18.559+01:00 level=INFO source=images.go:483 msg="total unused blobs removed: 0"
time=2025-10-26T09:40:18.559+01:00 level=INFO source=routes.go:1288 msg="Listening on 127.0.0.1:11434 (version 0.9.3)"
time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs"
time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu_windows.go:167 msg=packages count=1
time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu_windows.go:183 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu_windows.go:214 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:40:20.336+01:00 level=INFO source=gpu.go:319 msg="detected OS VRAM overhead" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" overhead="894.0 MiB"
time=2025-10-26T09:40:20.426+01:00 level=INFO source=amd_hip_windows.go:103 msg="AMD ROCm reports no devices found"
time=2025-10-26T09:40:20.426+01:00 level=INFO source=amd_windows.go:49 msg="no compatible amdgpu devices detected"
time=2025-10-26T09:40:20.428+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda variant=v12 compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" total="8.0 GiB" available="6.9 GiB"
[GIN] 2025/10/26 - 09:40:20 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 09:40:20 | 200 |      3.1617ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/26 - 09:40:34 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 09:40:34 | 200 |     30.8289ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T09:40:34.774+01:00 level=INFO source=server.go:135 msg="system memory" total="31.1 GiB" free="20.8 GiB" free_swap="49.9 GiB"
time=2025-10-26T09:40:34.774+01:00 level=INFO source=server.go:175 msg=offload library=cuda layers.requested=-1 layers.model=49 layers.offload=21 layers.split="" memory.available="[6.9 GiB]" memory.gpu_overhead="0 B" memory.required.full="13.4 GiB" memory.required.partial="6.7 GiB" memory.required.kv="384.0 MiB" memory.required.allocations="[6.7 GiB]" memory.weights.total="11.8 GiB" memory.weights.repeating="11.6 GiB" memory.weights.nonrepeating="243.4 MiB" memory.graph.full="512.0 MiB" memory.graph.partial="512.0 MiB"
llama_model_loader: loaded meta data with 47 key-value pairs and 579 tensors from C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = qwen3moe
llama_model_loader: - kv   1:                               general.type str              = model
llama_model_loader: - kv   2:                               general.name str              = Qwen3-Coder-30B-A3B-Instruct-1M
llama_model_loader: - kv   3:                           general.finetune str              = Instruct-1m
llama_model_loader: - kv   4:                           general.basename str              = Qwen3-Coder-30B-A3B-Instruct-1M
llama_model_loader: - kv   5:                       general.quantized_by str              = Unsloth
llama_model_loader: - kv   6:                         general.size_label str              = 30B-A3B
llama_model_loader: - kv   7:                            general.license str              = apache-2.0
llama_model_loader: - kv   8:                       general.license.link str              = https://huggingface.co/Qwen/Qwen3-Cod...
llama_model_loader: - kv   9:                           general.repo_url str              = https://huggingface.co/unsloth
llama_model_loader: - kv  10:                   general.base_model.count u32              = 1
llama_model_loader: - kv  11:                  general.base_model.0.name str              = Qwen3 Coder 30B A3B Instruct
llama_model_loader: - kv  12:          general.base_model.0.organization str              = Qwen
llama_model_loader: - kv  13:              general.base_model.0.repo_url str              = https://huggingface.co/Qwen/Qwen3-Cod...
llama_model_loader: - kv  14:                               general.tags arr[str,2]       = ["unsloth", "text-generation"]
llama_model_loader: - kv  15:                       qwen3moe.block_count u32              = 48
llama_model_loader: - kv  16:                    qwen3moe.context_length u32              = 1048576
llama_model_loader: - kv  17:                  qwen3moe.embedding_length u32              = 2048
llama_model_loader: - kv  18:               qwen3moe.feed_forward_length u32              = 5472
llama_model_loader: - kv  19:              qwen3moe.attention.head_count u32              = 32
llama_model_loader: - kv  20:           qwen3moe.attention.head_count_kv u32              = 4
llama_model_loader: - kv  21:                    qwen3moe.rope.freq_base f32              = 10000000.000000
llama_model_loader: - kv  22:  qwen3moe.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv  23:                 qwen3moe.expert_used_count u32              = 8
llama_model_loader: - kv  24:              qwen3moe.attention.key_length u32              = 128
llama_model_loader: - kv  25:            qwen3moe.attention.value_length u32              = 128
llama_model_loader: - kv  26:                      qwen3moe.expert_count u32              = 128
llama_model_loader: - kv  27:        qwen3moe.expert_feed_forward_length u32              = 768
llama_model_loader: - kv  28: qwen3moe.expert_shared_feed_forward_length u32              = 0
llama_model_loader: - kv  29:                 qwen3moe.rope.scaling.type str              = yarn
llama_model_loader: - kv  30:               qwen3moe.rope.scaling.factor f32              = 4.000000
llama_model_loader: - kv  31: qwen3moe.rope.scaling.original_context_length u32              = 262144
llama_model_loader: - kv  32:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  33:                         tokenizer.ggml.pre str              = qwen2
llama_model_loader: - kv  34:                      tokenizer.ggml.tokens arr[str,151936]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  35:                  tokenizer.ggml.token_type arr[i32,151936]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  36:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv  37:                tokenizer.ggml.eos_token_id u32              = 151645
llama_model_loader: - kv  38:            tokenizer.ggml.padding_token_id u32              = 151654
llama_model_loader: - kv  39:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  40:                    tokenizer.chat_template str              = {# Copyright 2025-present Unsloth. Ap...
llama_model_loader: - kv  41:               general.quantization_version u32              = 2
llama_model_loader: - kv  42:                          general.file_type u32              = 23
llama_model_loader: - kv  43:                      quantize.imatrix.file str              = Qwen3-Coder-30B-A3B-Instruct-1M-GGUF/...
llama_model_loader: - kv  44:                   quantize.imatrix.dataset str              = unsloth_calibration_Qwen3-Coder-30B-A...
llama_model_loader: - kv  45:             quantize.imatrix.entries_count u32              = 384
llama_model_loader: - kv  46:              quantize.imatrix.chunks_count u32              = 154
llama_model_loader: - type  f32:  241 tensors
llama_model_loader: - type q4_K:    1 tensors
llama_model_loader: - type q5_K:   18 tensors
llama_model_loader: - type q6_K:   11 tensors
llama_model_loader: - type iq3_xxs:   91 tensors
llama_model_loader: - type iq3_s:   40 tensors
llama_model_loader: - type iq4_xs:  177 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type   = IQ3_XXS - 3.0625 bpw
print_info: file size   = 11.98 GiB (3.37 BPW) 
load: special tokens cache size = 26
load: token to piece cache size = 0.9311 MB
print_info: arch             = qwen3moe
print_info: vocab_only       = 1
print_info: model type       = ?B
print_info: model params     = 30.53 B
print_info: general.name     = Qwen3-Coder-30B-A3B-Instruct-1M
print_info: n_ff_exp         = 0
print_info: vocab type       = BPE
print_info: n_vocab          = 151936
print_info: n_merges         = 151387
print_info: BOS token        = 11 ','
print_info: EOS token        = 151645 '<|im_end|>'
print_info: EOT token        = 151645 '<|im_end|>'
print_info: PAD token        = 151654 '<|vision_pad|>'
print_info: LF token         = 198 'Ċ'
print_info: FIM PRE token    = 151659 '<|fim_prefix|>'
print_info: FIM SUF token    = 151661 '<|fim_suffix|>'
print_info: FIM MID token    = 151660 '<|fim_middle|>'
print_info: FIM PAD token    = 151662 '<|fim_pad|>'
print_info: FIM REP token    = 151663 '<|repo_name|>'
print_info: FIM SEP token    = 151664 '<|file_sep|>'
print_info: EOG token        = 151643 '<|endoftext|>'
print_info: EOG token        = 151645 '<|im_end|>'
print_info: EOG token        = 151662 '<|fim_pad|>'
print_info: EOG token        = 151663 '<|repo_name|>'
print_info: EOG token        = 151664 '<|file_sep|>'
print_info: max token length = 256
llama_model_load: vocab only - skipping tensors
time=2025-10-26T09:40:34.907+01:00 level=INFO source=server.go:438 msg="starting llama server" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --ctx-size 4096 --batch-size 512 --n-gpu-layers 21 --threads 4 --no-mmap --parallel 1 --port 55530"
time=2025-10-26T09:40:34.910+01:00 level=INFO source=sched.go:483 msg="loaded runners" count=1
time=2025-10-26T09:40:34.910+01:00 level=INFO source=server.go:598 msg="waiting for llama runner to start responding"
time=2025-10-26T09:40:34.910+01:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server error"
time=2025-10-26T09:40:34.943+01:00 level=INFO source=runner.go:815 msg="starting go runner"
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cuda.dll
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
time=2025-10-26T09:40:35.070+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T09:40:35.071+01:00 level=INFO source=runner.go:874 msg="Server listening on 127.0.0.1:55530"
llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce RTX 4070 Laptop GPU) - 7054 MiB free
llama_model_loader: loaded meta data with 47 key-value pairs and 579 tensors from C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = qwen3moe
llama_model_loader: - kv   1:                               general.type str              = model
llama_model_loader: - kv   2:                               general.name str              = Qwen3-Coder-30B-A3B-Instruct-1M
llama_model_loader: - kv   3:                           general.finetune str              = Instruct-1m
llama_model_loader: - kv   4:                           general.basename str              = Qwen3-Coder-30B-A3B-Instruct-1M
llama_model_loader: - kv   5:                       general.quantized_by str              = Unsloth
llama_model_loader: - kv   6:                         general.size_label str              = 30B-A3B
llama_model_loader: - kv   7:                            general.license str              = apache-2.0
llama_model_loader: - kv   8:                       general.license.link str              = https://huggingface.co/Qwen/Qwen3-Cod...
llama_model_loader: - kv   9:                           general.repo_url str              = https://huggingface.co/unsloth
llama_model_loader: - kv  10:                   general.base_model.count u32              = 1
llama_model_loader: - kv  11:                  general.base_model.0.name str              = Qwen3 Coder 30B A3B Instruct
llama_model_loader: - kv  12:          general.base_model.0.organization str              = Qwen
llama_model_loader: - kv  13:              general.base_model.0.repo_url str              = https://huggingface.co/Qwen/Qwen3-Cod...
llama_model_loader: - kv  14:                               general.tags arr[str,2]       = ["unsloth", "text-generation"]
llama_model_loader: - kv  15:                       qwen3moe.block_count u32              = 48
llama_model_loader: - kv  16:                    qwen3moe.context_length u32              = 1048576
llama_model_loader: - kv  17:                  qwen3moe.embedding_length u32              = 2048
llama_model_loader: - kv  18:               qwen3moe.feed_forward_length u32              = 5472
llama_model_loader: - kv  19:              qwen3moe.attention.head_count u32              = 32
llama_model_loader: - kv  20:           qwen3moe.attention.head_count_kv u32              = 4
llama_model_loader: - kv  21:                    qwen3moe.rope.freq_base f32              = 10000000.000000
llama_model_loader: - kv  22:  qwen3moe.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv  23:                 qwen3moe.expert_used_count u32              = 8
llama_model_loader: - kv  24:              qwen3moe.attention.key_length u32              = 128
llama_model_loader: - kv  25:            qwen3moe.attention.value_length u32              = 128
llama_model_loader: - kv  26:                      qwen3moe.expert_count u32              = 128
llama_model_loader: - kv  27:        qwen3moe.expert_feed_forward_length u32              = 768
llama_model_loader: - kv  28: qwen3moe.expert_shared_feed_forward_length u32              = 0
llama_model_loader: - kv  29:                 qwen3moe.rope.scaling.type str              = yarn
llama_model_loader: - kv  30:               qwen3moe.rope.scaling.factor f32              = 4.000000
llama_model_loader: - kv  31: qwen3moe.rope.scaling.original_context_length u32              = 262144
llama_model_loader: - kv  32:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  33:                         tokenizer.ggml.pre str              = qwen2
llama_model_loader: - kv  34:                      tokenizer.ggml.tokens arr[str,151936]  = ["!", "\"", "#", "$", "%", "&", "'", ...
time=2025-10-26T09:40:35.162+01:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server loading model"
llama_model_loader: - kv  35:                  tokenizer.ggml.token_type arr[i32,151936]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  36:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv  37:                tokenizer.ggml.eos_token_id u32              = 151645
llama_model_loader: - kv  38:            tokenizer.ggml.padding_token_id u32              = 151654
llama_model_loader: - kv  39:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  40:                    tokenizer.chat_template str              = {# Copyright 2025-present Unsloth. Ap...
llama_model_loader: - kv  41:               general.quantization_version u32              = 2
llama_model_loader: - kv  42:                          general.file_type u32              = 23
llama_model_loader: - kv  43:                      quantize.imatrix.file str              = Qwen3-Coder-30B-A3B-Instruct-1M-GGUF/...
llama_model_loader: - kv  44:                   quantize.imatrix.dataset str              = unsloth_calibration_Qwen3-Coder-30B-A...
llama_model_loader: - kv  45:             quantize.imatrix.entries_count u32              = 384
llama_model_loader: - kv  46:              quantize.imatrix.chunks_count u32              = 154
llama_model_loader: - type  f32:  241 tensors
llama_model_loader: - type q4_K:    1 tensors
llama_model_loader: - type q5_K:   18 tensors
llama_model_loader: - type q6_K:   11 tensors
llama_model_loader: - type iq3_xxs:   91 tensors
llama_model_loader: - type iq3_s:   40 tensors
llama_model_loader: - type iq4_xs:  177 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type   = IQ3_XXS - 3.0625 bpw
print_info: file size   = 11.98 GiB (3.37 BPW) 
load: special tokens cache size = 26
load: token to piece cache size = 0.9311 MB
print_info: arch             = qwen3moe
print_info: vocab_only       = 0
print_info: n_ctx_train      = 1048576
print_info: n_embd           = 2048
print_info: n_layer          = 48
print_info: n_head           = 32
print_info: n_head_kv        = 4
print_info: n_rot            = 128
print_info: n_swa            = 0
print_info: n_swa_pattern    = 1
print_info: n_embd_head_k    = 128
print_info: n_embd_head_v    = 128
print_info: n_gqa            = 8
print_info: n_embd_k_gqa     = 512
print_info: n_embd_v_gqa     = 512
print_info: f_norm_eps       = 0.0e+00
print_info: f_norm_rms_eps   = 1.0e-06
print_info: f_clamp_kqv      = 0.0e+00
print_info: f_max_alibi_bias = 0.0e+00
print_info: f_logit_scale    = 0.0e+00
print_info: f_attn_scale     = 0.0e+00
print_info: n_ff             = 5472
print_info: n_expert         = 128
print_info: n_expert_used    = 8
print_info: causal attn      = 1
print_info: pooling type     = 0
print_info: rope type        = 2
print_info: rope scaling     = yarn
print_info: freq_base_train  = 10000000.0
print_info: freq_scale_train = 0.25
print_info: n_ctx_orig_yarn  = 262144
print_info: rope_finetuned   = unknown
print_info: ssm_d_conv       = 0
print_info: ssm_d_inner      = 0
print_info: ssm_d_state      = 0
print_info: ssm_dt_rank      = 0
print_info: ssm_dt_b_c_rms   = 0
print_info: model type       = 30B.A3B
print_info: model params     = 30.53 B
print_info: general.name     = Qwen3-Coder-30B-A3B-Instruct-1M
print_info: n_ff_exp         = 768
print_info: vocab type       = BPE
print_info: n_vocab          = 151936
print_info: n_merges         = 151387
print_info: BOS token        = 11 ','
print_info: EOS token        = 151645 '<|im_end|>'
print_info: EOT token        = 151645 '<|im_end|>'
print_info: PAD token        = 151654 '<|vision_pad|>'
print_info: LF token         = 198 'Ċ'
print_info: FIM PRE token    = 151659 '<|fim_prefix|>'
print_info: FIM SUF token    = 151661 '<|fim_suffix|>'
print_info: FIM MID token    = 151660 '<|fim_middle|>'
print_info: FIM PAD token    = 151662 '<|fim_pad|>'
print_info: FIM REP token    = 151663 '<|repo_name|>'
print_info: FIM SEP token    = 151664 '<|file_sep|>'
print_info: EOG token        = 151643 '<|endoftext|>'
print_info: EOG token        = 151645 '<|im_end|>'
print_info: EOG token        = 151662 '<|fim_pad|>'
print_info: EOG token        = 151663 '<|repo_name|>'
print_info: EOG token        = 151664 '<|file_sep|>'
print_info: max token length = 256
load_tensors: loading model tensors, this can take a while... (mmap = false)
load_tensors: offloading 21 repeating layers to GPU
load_tensors: offloaded 21/49 layers to GPU
load_tensors:    CUDA_Host model buffer size =  6654.13 MiB
load_tensors:        CUDA0 model buffer size =  5446.29 MiB
load_tensors:          CPU model buffer size =   166.92 MiB
llama_context: constructing llama_context
llama_context: n_seq_max     = 1
llama_context: n_ctx         = 4096
llama_context: n_ctx_per_seq = 4096
llama_context: n_batch       = 512
llama_context: n_ubatch      = 512
llama_context: causal_attn   = 1
llama_context: flash_attn    = 0
llama_context: freq_base     = 10000000.0
llama_context: freq_scale    = 0.25
llama_context: n_ctx_per_seq (4096) < n_ctx_train (1048576) -- the full capacity of the model will not be utilized
llama_context:        CPU  output buffer size =     0.59 MiB
llama_kv_cache_unified: kv_size = 4096, type_k = 'f16', type_v = 'f16', n_layer = 48, can_shift = 1, padding = 32
llama_kv_cache_unified:      CUDA0 KV buffer size =   168.00 MiB
llama_kv_cache_unified:        CPU KV buffer size =   216.00 MiB
llama_kv_cache_unified: KV self size  =  384.00 MiB, K (f16):  192.00 MiB, V (f16):  192.00 MiB
llama_context:      CUDA0 compute buffer size =   544.18 MiB
llama_context:  CUDA_Host compute buffer size =    12.01 MiB
llama_context: graph nodes  = 3126
llama_context: graph splits = 382 (with bs=512), 57 (with bs=1)
time=2025-10-26T09:40:46.681+01:00 level=INFO source=server.go:637 msg="llama runner started in 11.77 seconds"
[GIN] 2025/10/26 - 09:40:46 | 200 |   11.9995817s |       127.0.0.1 | POST     "/api/generate"
[GIN] 2025/10/26 - 09:41:32 | 200 |    23.624138s |       127.0.0.1 | POST     "/api/chat"
time=2025-10-26T09:41:33.581+01:00 level=INFO source=routes.go:1235 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-26T09:41:33.611+01:00 level=INFO source=images.go:476 msg="total blobs: 32"
time=2025-10-26T09:41:33.614+01:00 level=INFO source=images.go:483 msg="total unused blobs removed: 0"
time=2025-10-26T09:41:33.616+01:00 level=INFO source=routes.go:1288 msg="Listening on 127.0.0.1:11434 (version 0.9.3)"
time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs"
time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu_windows.go:167 msg=packages count=1
time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu_windows.go:183 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu_windows.go:214 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:41:33.753+01:00 level=INFO source=gpu.go:319 msg="detected OS VRAM overhead" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" overhead="592.2 MiB"
time=2025-10-26T09:41:33.807+01:00 level=INFO source=amd_hip_windows.go:103 msg="AMD ROCm reports no devices found"
time=2025-10-26T09:41:33.807+01:00 level=INFO source=amd_windows.go:49 msg="no compatible amdgpu devices detected"
time=2025-10-26T09:41:33.808+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda variant=v12 compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" total="8.0 GiB" available="6.9 GiB"
  • now finally for those 5% rare cases when ollama 0.12.6 would run through:
time=2025-10-26T09:59:21.323+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-26T09:59:21.344+01:00 level=INFO source=images.go:522 msg="total blobs: 32"
time=2025-10-26T09:59:21.346+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0"
time=2025-10-26T09:59:21.347+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)"
time=2025-10-26T09:59:21.349+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..."
time=2025-10-26T09:59:24.179+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v13 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB"
time=2025-10-26T09:59:24.179+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB"
[GIN] 2025/10/26 - 09:59:24 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 09:59:24 | 200 |      2.0815ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/26 - 09:59:34 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 09:59:34 | 200 |     30.6683ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T09:59:34.528+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:34.528+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:34.528+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:34.528+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T09:59:34.528+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 64105"
time=2025-10-26T09:59:34.533+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T09:59:34.533+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:34.533+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:34.533+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:34.533+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="21.0 GiB" free_swap="49.8 GiB"
time=2025-10-26T09:59:34.533+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T09:59:34.568+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T09:59:34.587+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:64105"
time=2025-10-26T09:59:34.588+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:34.598+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T09:59:34.690+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T09:59:34.927+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:34.947+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T09:59:35.367+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:64105/load\": read tcp 127.0.0.1:64111->127.0.0.1:64105: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 09:59:35 | 500 |    1.1223942s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T09:59:35.396+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 09:59:40 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 09:59:40 | 200 |     30.0767ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 64131"
time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.8 GiB" free_swap="48.6 GiB"
time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T09:59:41.411+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T09:59:41.427+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:64131"
time=2025-10-26T09:59:41.429+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:41.441+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T09:59:41.527+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T09:59:41.729+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:41.751+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T09:59:42.130+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:64131/load\": read tcp 127.0.0.1:64137->127.0.0.1:64131: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 09:59:42 | 500 |    1.5338729s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T09:59:42.159+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 09:59:49 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 09:59:49 | 200 |     30.9264ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T09:59:50.206+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:50.206+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:50.206+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:50.206+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T09:59:50.207+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 64176"
time=2025-10-26T09:59:50.210+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T09:59:50.210+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:50.210+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:50.210+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:50.210+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="18.7 GiB" free_swap="45.5 GiB"
time=2025-10-26T09:59:50.210+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T09:59:50.244+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T09:59:50.260+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:64176"
time=2025-10-26T09:59:50.266+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:50.281+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T09:59:50.453+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T09:59:50.668+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:50.692+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T09:59:51.074+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:64176/load\": read tcp 127.0.0.1:64182->127.0.0.1:64176: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 09:59:51 | 500 |    1.1232683s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T09:59:51.101+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 09:59:57 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 09:59:57 | 200 |     29.2615ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T09:59:57.355+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:57.355+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:57.355+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:57.355+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T09:59:57.356+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 57432"
time=2025-10-26T09:59:57.361+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T09:59:57.361+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T09:59:57.361+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T09:59:57.361+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T09:59:57.361+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.1 GiB" free_swap="43.0 GiB"
time=2025-10-26T09:59:57.361+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.0 GiB" free="7.5 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T09:59:57.402+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T09:59:57.419+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:57432"
time=2025-10-26T09:59:57.428+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:57.444+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T09:59:57.525+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T09:59:57.737+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:26[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:26(22..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:57.759+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:26[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:26(22..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:26[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:26(22..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=ggml.go:480 msg="offloading 26 repeating layers to GPU"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=ggml.go:484 msg="offloading output layer to CPU"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=ggml.go:492 msg="offloaded 26/49 layers to GPU"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.5 GiB"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.5 GiB"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="208.0 MiB"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="176.0 MiB"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T09:59:58.000+01:00 level=INFO source=sched.go:482 msg="loaded runners" count=1
time=2025-10-26T09:59:58.000+01:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding"
time=2025-10-26T09:59:58.001+01:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model"
time=2025-10-26T10:00:11.780+01:00 level=INFO source=server.go:1310 msg="llama runner started in 14.42 seconds"
[GIN] 2025/10/26 - 10:00:11 | 200 |   14.7022691s |       127.0.0.1 | POST     "/api/generate"
[GIN] 2025/10/26 - 10:00:26 | 200 |            0s |       127.0.0.1 | GET      "/api/version"
[GIN] 2025/10/26 - 10:00:26 | 200 |       529.6µs |       127.0.0.1 | GET      "/api/version"
[GIN] 2025/10/26 - 10:00:26 | 200 |      7.3573ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/26 - 10:00:26 | 404 |      2.1446ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 10:00:27 | 404 |       3.115ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 10:00:29 | 404 |      2.1104ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 10:01:20 | 200 |   47.8262204s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2025/10/26 - 10:01:49 | 200 |            0s |       127.0.0.1 | GET      "/api/version"
<!-- gh-comment-id:3448704765 --> @kiliansinger commented on GitHub (Oct 26, 2025): Here is the server log starting ```bash ollama run danielsheep/Qwen3-Coder-30B-A3B-Instruct-1M-Unsloth:UD-IQ3_XXS ``` in the following presented for ollama 0.12.6 (failing approx. 95% of cases), then for ollama 0.9.3 (running stable 100% of cases). Interestingly I also found that in 5% of cases (after rebooting) ollama 0.12.6 would run through. This is also listed as third log. Here are the log outputs * with ollama 0.12.6 failing: ``` time=2025-10-26T17:52:03.657+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-26T17:52:03.677+01:00 level=INFO source=images.go:522 msg="total blobs: 32" time=2025-10-26T17:52:03.679+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0" time=2025-10-26T17:52:03.680+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)" time=2025-10-26T17:52:03.680+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..." time=2025-10-26T17:52:05.280+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v13 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB" time=2025-10-26T17:52:05.281+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB" [GIN] 2025/10/26 - 17:52:05 | 200 | 562µs | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 17:52:05 | 200 | 2.2054ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/26 - 17:52:36 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 17:52:36 | 500 | 0s | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 17:52:45 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 17:52:45 | 200 | 32.3612ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T17:52:46.148+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T17:52:46.148+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T17:52:46.148+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T17:52:46.148+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T17:52:46.149+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50809" time=2025-10-26T17:52:46.152+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T17:52:46.152+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T17:52:46.152+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T17:52:46.152+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T17:52:46.152+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.3 GiB" time=2025-10-26T17:52:46.152+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T17:52:46.194+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T17:52:46.211+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50809" time=2025-10-26T17:52:46.220+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T17:52:46.236+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T17:52:46.320+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T17:52:46.548+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T17:52:46.570+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T17:52:46.969+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T17:52:46.970+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T17:52:46.970+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50809/load\": read tcp 127.0.0.1:50816->127.0.0.1:50809: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 17:52:46 | 500 | 1.1105028s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T17:52:46.995+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" ``` * with ollama 0.9.3 running through 100% reliable ``` time=2025-10-26T09:40:18.531+01:00 level=INFO source=routes.go:1235 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-26T09:40:18.553+01:00 level=INFO source=images.go:476 msg="total blobs: 32" time=2025-10-26T09:40:18.559+01:00 level=INFO source=images.go:483 msg="total unused blobs removed: 0" time=2025-10-26T09:40:18.559+01:00 level=INFO source=routes.go:1288 msg="Listening on 127.0.0.1:11434 (version 0.9.3)" time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs" time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu_windows.go:167 msg=packages count=1 time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu_windows.go:183 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:40:18.559+01:00 level=INFO source=gpu_windows.go:214 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:40:20.336+01:00 level=INFO source=gpu.go:319 msg="detected OS VRAM overhead" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" overhead="894.0 MiB" time=2025-10-26T09:40:20.426+01:00 level=INFO source=amd_hip_windows.go:103 msg="AMD ROCm reports no devices found" time=2025-10-26T09:40:20.426+01:00 level=INFO source=amd_windows.go:49 msg="no compatible amdgpu devices detected" time=2025-10-26T09:40:20.428+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda variant=v12 compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" total="8.0 GiB" available="6.9 GiB" [GIN] 2025/10/26 - 09:40:20 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 09:40:20 | 200 | 3.1617ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/26 - 09:40:34 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 09:40:34 | 200 | 30.8289ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T09:40:34.774+01:00 level=INFO source=server.go:135 msg="system memory" total="31.1 GiB" free="20.8 GiB" free_swap="49.9 GiB" time=2025-10-26T09:40:34.774+01:00 level=INFO source=server.go:175 msg=offload library=cuda layers.requested=-1 layers.model=49 layers.offload=21 layers.split="" memory.available="[6.9 GiB]" memory.gpu_overhead="0 B" memory.required.full="13.4 GiB" memory.required.partial="6.7 GiB" memory.required.kv="384.0 MiB" memory.required.allocations="[6.7 GiB]" memory.weights.total="11.8 GiB" memory.weights.repeating="11.6 GiB" memory.weights.nonrepeating="243.4 MiB" memory.graph.full="512.0 MiB" memory.graph.partial="512.0 MiB" llama_model_loader: loaded meta data with 47 key-value pairs and 579 tensors from C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen3moe llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen3-Coder-30B-A3B-Instruct-1M llama_model_loader: - kv 3: general.finetune str = Instruct-1m llama_model_loader: - kv 4: general.basename str = Qwen3-Coder-30B-A3B-Instruct-1M llama_model_loader: - kv 5: general.quantized_by str = Unsloth llama_model_loader: - kv 6: general.size_label str = 30B-A3B llama_model_loader: - kv 7: general.license str = apache-2.0 llama_model_loader: - kv 8: general.license.link str = https://huggingface.co/Qwen/Qwen3-Cod... llama_model_loader: - kv 9: general.repo_url str = https://huggingface.co/unsloth llama_model_loader: - kv 10: general.base_model.count u32 = 1 llama_model_loader: - kv 11: general.base_model.0.name str = Qwen3 Coder 30B A3B Instruct llama_model_loader: - kv 12: general.base_model.0.organization str = Qwen llama_model_loader: - kv 13: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen3-Cod... llama_model_loader: - kv 14: general.tags arr[str,2] = ["unsloth", "text-generation"] llama_model_loader: - kv 15: qwen3moe.block_count u32 = 48 llama_model_loader: - kv 16: qwen3moe.context_length u32 = 1048576 llama_model_loader: - kv 17: qwen3moe.embedding_length u32 = 2048 llama_model_loader: - kv 18: qwen3moe.feed_forward_length u32 = 5472 llama_model_loader: - kv 19: qwen3moe.attention.head_count u32 = 32 llama_model_loader: - kv 20: qwen3moe.attention.head_count_kv u32 = 4 llama_model_loader: - kv 21: qwen3moe.rope.freq_base f32 = 10000000.000000 llama_model_loader: - kv 22: qwen3moe.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 23: qwen3moe.expert_used_count u32 = 8 llama_model_loader: - kv 24: qwen3moe.attention.key_length u32 = 128 llama_model_loader: - kv 25: qwen3moe.attention.value_length u32 = 128 llama_model_loader: - kv 26: qwen3moe.expert_count u32 = 128 llama_model_loader: - kv 27: qwen3moe.expert_feed_forward_length u32 = 768 llama_model_loader: - kv 28: qwen3moe.expert_shared_feed_forward_length u32 = 0 llama_model_loader: - kv 29: qwen3moe.rope.scaling.type str = yarn llama_model_loader: - kv 30: qwen3moe.rope.scaling.factor f32 = 4.000000 llama_model_loader: - kv 31: qwen3moe.rope.scaling.original_context_length u32 = 262144 llama_model_loader: - kv 32: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 33: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 34: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 35: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 36: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 37: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 38: tokenizer.ggml.padding_token_id u32 = 151654 llama_model_loader: - kv 39: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 40: tokenizer.chat_template str = {# Copyright 2025-present Unsloth. Ap... llama_model_loader: - kv 41: general.quantization_version u32 = 2 llama_model_loader: - kv 42: general.file_type u32 = 23 llama_model_loader: - kv 43: quantize.imatrix.file str = Qwen3-Coder-30B-A3B-Instruct-1M-GGUF/... llama_model_loader: - kv 44: quantize.imatrix.dataset str = unsloth_calibration_Qwen3-Coder-30B-A... llama_model_loader: - kv 45: quantize.imatrix.entries_count u32 = 384 llama_model_loader: - kv 46: quantize.imatrix.chunks_count u32 = 154 llama_model_loader: - type f32: 241 tensors llama_model_loader: - type q4_K: 1 tensors llama_model_loader: - type q5_K: 18 tensors llama_model_loader: - type q6_K: 11 tensors llama_model_loader: - type iq3_xxs: 91 tensors llama_model_loader: - type iq3_s: 40 tensors llama_model_loader: - type iq4_xs: 177 tensors print_info: file format = GGUF V3 (latest) print_info: file type = IQ3_XXS - 3.0625 bpw print_info: file size = 11.98 GiB (3.37 BPW) load: special tokens cache size = 26 load: token to piece cache size = 0.9311 MB print_info: arch = qwen3moe print_info: vocab_only = 1 print_info: model type = ?B print_info: model params = 30.53 B print_info: general.name = Qwen3-Coder-30B-A3B-Instruct-1M print_info: n_ff_exp = 0 print_info: vocab type = BPE print_info: n_vocab = 151936 print_info: n_merges = 151387 print_info: BOS token = 11 ',' print_info: EOS token = 151645 '<|im_end|>' print_info: EOT token = 151645 '<|im_end|>' print_info: PAD token = 151654 '<|vision_pad|>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 151643 '<|endoftext|>' print_info: EOG token = 151645 '<|im_end|>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 llama_model_load: vocab only - skipping tensors time=2025-10-26T09:40:34.907+01:00 level=INFO source=server.go:438 msg="starting llama server" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --ctx-size 4096 --batch-size 512 --n-gpu-layers 21 --threads 4 --no-mmap --parallel 1 --port 55530" time=2025-10-26T09:40:34.910+01:00 level=INFO source=sched.go:483 msg="loaded runners" count=1 time=2025-10-26T09:40:34.910+01:00 level=INFO source=server.go:598 msg="waiting for llama runner to start responding" time=2025-10-26T09:40:34.910+01:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server error" time=2025-10-26T09:40:34.943+01:00 level=INFO source=runner.go:815 msg="starting go runner" ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cuda.dll load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll time=2025-10-26T09:40:35.070+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T09:40:35.071+01:00 level=INFO source=runner.go:874 msg="Server listening on 127.0.0.1:55530" llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce RTX 4070 Laptop GPU) - 7054 MiB free llama_model_loader: loaded meta data with 47 key-value pairs and 579 tensors from C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = qwen3moe llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen3-Coder-30B-A3B-Instruct-1M llama_model_loader: - kv 3: general.finetune str = Instruct-1m llama_model_loader: - kv 4: general.basename str = Qwen3-Coder-30B-A3B-Instruct-1M llama_model_loader: - kv 5: general.quantized_by str = Unsloth llama_model_loader: - kv 6: general.size_label str = 30B-A3B llama_model_loader: - kv 7: general.license str = apache-2.0 llama_model_loader: - kv 8: general.license.link str = https://huggingface.co/Qwen/Qwen3-Cod... llama_model_loader: - kv 9: general.repo_url str = https://huggingface.co/unsloth llama_model_loader: - kv 10: general.base_model.count u32 = 1 llama_model_loader: - kv 11: general.base_model.0.name str = Qwen3 Coder 30B A3B Instruct llama_model_loader: - kv 12: general.base_model.0.organization str = Qwen llama_model_loader: - kv 13: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen3-Cod... llama_model_loader: - kv 14: general.tags arr[str,2] = ["unsloth", "text-generation"] llama_model_loader: - kv 15: qwen3moe.block_count u32 = 48 llama_model_loader: - kv 16: qwen3moe.context_length u32 = 1048576 llama_model_loader: - kv 17: qwen3moe.embedding_length u32 = 2048 llama_model_loader: - kv 18: qwen3moe.feed_forward_length u32 = 5472 llama_model_loader: - kv 19: qwen3moe.attention.head_count u32 = 32 llama_model_loader: - kv 20: qwen3moe.attention.head_count_kv u32 = 4 llama_model_loader: - kv 21: qwen3moe.rope.freq_base f32 = 10000000.000000 llama_model_loader: - kv 22: qwen3moe.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 23: qwen3moe.expert_used_count u32 = 8 llama_model_loader: - kv 24: qwen3moe.attention.key_length u32 = 128 llama_model_loader: - kv 25: qwen3moe.attention.value_length u32 = 128 llama_model_loader: - kv 26: qwen3moe.expert_count u32 = 128 llama_model_loader: - kv 27: qwen3moe.expert_feed_forward_length u32 = 768 llama_model_loader: - kv 28: qwen3moe.expert_shared_feed_forward_length u32 = 0 llama_model_loader: - kv 29: qwen3moe.rope.scaling.type str = yarn llama_model_loader: - kv 30: qwen3moe.rope.scaling.factor f32 = 4.000000 llama_model_loader: - kv 31: qwen3moe.rope.scaling.original_context_length u32 = 262144 llama_model_loader: - kv 32: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 33: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 34: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... time=2025-10-26T09:40:35.162+01:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server loading model" llama_model_loader: - kv 35: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 36: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 37: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 38: tokenizer.ggml.padding_token_id u32 = 151654 llama_model_loader: - kv 39: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 40: tokenizer.chat_template str = {# Copyright 2025-present Unsloth. Ap... llama_model_loader: - kv 41: general.quantization_version u32 = 2 llama_model_loader: - kv 42: general.file_type u32 = 23 llama_model_loader: - kv 43: quantize.imatrix.file str = Qwen3-Coder-30B-A3B-Instruct-1M-GGUF/... llama_model_loader: - kv 44: quantize.imatrix.dataset str = unsloth_calibration_Qwen3-Coder-30B-A... llama_model_loader: - kv 45: quantize.imatrix.entries_count u32 = 384 llama_model_loader: - kv 46: quantize.imatrix.chunks_count u32 = 154 llama_model_loader: - type f32: 241 tensors llama_model_loader: - type q4_K: 1 tensors llama_model_loader: - type q5_K: 18 tensors llama_model_loader: - type q6_K: 11 tensors llama_model_loader: - type iq3_xxs: 91 tensors llama_model_loader: - type iq3_s: 40 tensors llama_model_loader: - type iq4_xs: 177 tensors print_info: file format = GGUF V3 (latest) print_info: file type = IQ3_XXS - 3.0625 bpw print_info: file size = 11.98 GiB (3.37 BPW) load: special tokens cache size = 26 load: token to piece cache size = 0.9311 MB print_info: arch = qwen3moe print_info: vocab_only = 0 print_info: n_ctx_train = 1048576 print_info: n_embd = 2048 print_info: n_layer = 48 print_info: n_head = 32 print_info: n_head_kv = 4 print_info: n_rot = 128 print_info: n_swa = 0 print_info: n_swa_pattern = 1 print_info: n_embd_head_k = 128 print_info: n_embd_head_v = 128 print_info: n_gqa = 8 print_info: n_embd_k_gqa = 512 print_info: n_embd_v_gqa = 512 print_info: f_norm_eps = 0.0e+00 print_info: f_norm_rms_eps = 1.0e-06 print_info: f_clamp_kqv = 0.0e+00 print_info: f_max_alibi_bias = 0.0e+00 print_info: f_logit_scale = 0.0e+00 print_info: f_attn_scale = 0.0e+00 print_info: n_ff = 5472 print_info: n_expert = 128 print_info: n_expert_used = 8 print_info: causal attn = 1 print_info: pooling type = 0 print_info: rope type = 2 print_info: rope scaling = yarn print_info: freq_base_train = 10000000.0 print_info: freq_scale_train = 0.25 print_info: n_ctx_orig_yarn = 262144 print_info: rope_finetuned = unknown print_info: ssm_d_conv = 0 print_info: ssm_d_inner = 0 print_info: ssm_d_state = 0 print_info: ssm_dt_rank = 0 print_info: ssm_dt_b_c_rms = 0 print_info: model type = 30B.A3B print_info: model params = 30.53 B print_info: general.name = Qwen3-Coder-30B-A3B-Instruct-1M print_info: n_ff_exp = 768 print_info: vocab type = BPE print_info: n_vocab = 151936 print_info: n_merges = 151387 print_info: BOS token = 11 ',' print_info: EOS token = 151645 '<|im_end|>' print_info: EOT token = 151645 '<|im_end|>' print_info: PAD token = 151654 '<|vision_pad|>' print_info: LF token = 198 'Ċ' print_info: FIM PRE token = 151659 '<|fim_prefix|>' print_info: FIM SUF token = 151661 '<|fim_suffix|>' print_info: FIM MID token = 151660 '<|fim_middle|>' print_info: FIM PAD token = 151662 '<|fim_pad|>' print_info: FIM REP token = 151663 '<|repo_name|>' print_info: FIM SEP token = 151664 '<|file_sep|>' print_info: EOG token = 151643 '<|endoftext|>' print_info: EOG token = 151645 '<|im_end|>' print_info: EOG token = 151662 '<|fim_pad|>' print_info: EOG token = 151663 '<|repo_name|>' print_info: EOG token = 151664 '<|file_sep|>' print_info: max token length = 256 load_tensors: loading model tensors, this can take a while... (mmap = false) load_tensors: offloading 21 repeating layers to GPU load_tensors: offloaded 21/49 layers to GPU load_tensors: CUDA_Host model buffer size = 6654.13 MiB load_tensors: CUDA0 model buffer size = 5446.29 MiB load_tensors: CPU model buffer size = 166.92 MiB llama_context: constructing llama_context llama_context: n_seq_max = 1 llama_context: n_ctx = 4096 llama_context: n_ctx_per_seq = 4096 llama_context: n_batch = 512 llama_context: n_ubatch = 512 llama_context: causal_attn = 1 llama_context: flash_attn = 0 llama_context: freq_base = 10000000.0 llama_context: freq_scale = 0.25 llama_context: n_ctx_per_seq (4096) < n_ctx_train (1048576) -- the full capacity of the model will not be utilized llama_context: CPU output buffer size = 0.59 MiB llama_kv_cache_unified: kv_size = 4096, type_k = 'f16', type_v = 'f16', n_layer = 48, can_shift = 1, padding = 32 llama_kv_cache_unified: CUDA0 KV buffer size = 168.00 MiB llama_kv_cache_unified: CPU KV buffer size = 216.00 MiB llama_kv_cache_unified: KV self size = 384.00 MiB, K (f16): 192.00 MiB, V (f16): 192.00 MiB llama_context: CUDA0 compute buffer size = 544.18 MiB llama_context: CUDA_Host compute buffer size = 12.01 MiB llama_context: graph nodes = 3126 llama_context: graph splits = 382 (with bs=512), 57 (with bs=1) time=2025-10-26T09:40:46.681+01:00 level=INFO source=server.go:637 msg="llama runner started in 11.77 seconds" [GIN] 2025/10/26 - 09:40:46 | 200 | 11.9995817s | 127.0.0.1 | POST "/api/generate" [GIN] 2025/10/26 - 09:41:32 | 200 | 23.624138s | 127.0.0.1 | POST "/api/chat" time=2025-10-26T09:41:33.581+01:00 level=INFO source=routes.go:1235 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-26T09:41:33.611+01:00 level=INFO source=images.go:476 msg="total blobs: 32" time=2025-10-26T09:41:33.614+01:00 level=INFO source=images.go:483 msg="total unused blobs removed: 0" time=2025-10-26T09:41:33.616+01:00 level=INFO source=routes.go:1288 msg="Listening on 127.0.0.1:11434 (version 0.9.3)" time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs" time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu_windows.go:167 msg=packages count=1 time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu_windows.go:183 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:41:33.616+01:00 level=INFO source=gpu_windows.go:214 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:41:33.753+01:00 level=INFO source=gpu.go:319 msg="detected OS VRAM overhead" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" overhead="592.2 MiB" time=2025-10-26T09:41:33.807+01:00 level=INFO source=amd_hip_windows.go:103 msg="AMD ROCm reports no devices found" time=2025-10-26T09:41:33.807+01:00 level=INFO source=amd_windows.go:49 msg="no compatible amdgpu devices detected" time=2025-10-26T09:41:33.808+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=cuda variant=v12 compute=8.9 driver=13.0 name="NVIDIA GeForce RTX 4070 Laptop GPU" total="8.0 GiB" available="6.9 GiB" ``` * now finally for those 5% rare cases when ollama 0.12.6 would run through: ``` time=2025-10-26T09:59:21.323+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-26T09:59:21.344+01:00 level=INFO source=images.go:522 msg="total blobs: 32" time=2025-10-26T09:59:21.346+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0" time=2025-10-26T09:59:21.347+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)" time=2025-10-26T09:59:21.349+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..." time=2025-10-26T09:59:24.179+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v13 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB" time=2025-10-26T09:59:24.179+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB" [GIN] 2025/10/26 - 09:59:24 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 09:59:24 | 200 | 2.0815ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/26 - 09:59:34 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 09:59:34 | 200 | 30.6683ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T09:59:34.528+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:34.528+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:34.528+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:34.528+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T09:59:34.528+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 64105" time=2025-10-26T09:59:34.533+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T09:59:34.533+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:34.533+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:34.533+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:34.533+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="21.0 GiB" free_swap="49.8 GiB" time=2025-10-26T09:59:34.533+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T09:59:34.568+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T09:59:34.587+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:64105" time=2025-10-26T09:59:34.588+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:34.598+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T09:59:34.690+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T09:59:34.927+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:34.947+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T09:59:35.367+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T09:59:35.367+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:64105/load\": read tcp 127.0.0.1:64111->127.0.0.1:64105: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 09:59:35 | 500 | 1.1223942s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T09:59:35.396+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 09:59:40 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 09:59:40 | 200 | 30.0767ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 64131" time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:41.372+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.8 GiB" free_swap="48.6 GiB" time=2025-10-26T09:59:41.372+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T09:59:41.411+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T09:59:41.427+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:64131" time=2025-10-26T09:59:41.429+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:41.441+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T09:59:41.527+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T09:59:41.729+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:41.751+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T09:59:42.130+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T09:59:42.130+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:64131/load\": read tcp 127.0.0.1:64137->127.0.0.1:64131: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 09:59:42 | 500 | 1.5338729s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T09:59:42.159+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 09:59:49 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 09:59:49 | 200 | 30.9264ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T09:59:50.206+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:50.206+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:50.206+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:50.206+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T09:59:50.207+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 64176" time=2025-10-26T09:59:50.210+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T09:59:50.210+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:50.210+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:50.210+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:50.210+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="18.7 GiB" free_swap="45.5 GiB" time=2025-10-26T09:59:50.210+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T09:59:50.244+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T09:59:50.260+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:64176" time=2025-10-26T09:59:50.266+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:50.281+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T09:59:50.453+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T09:59:50.668+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:50.692+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T09:59:51.074+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T09:59:51.074+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:64176/load\": read tcp 127.0.0.1:64182->127.0.0.1:64176: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 09:59:51 | 500 | 1.1232683s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T09:59:51.101+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 09:59:57 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 09:59:57 | 200 | 29.2615ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T09:59:57.355+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:57.355+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:57.355+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:57.355+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T09:59:57.356+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 57432" time=2025-10-26T09:59:57.361+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T09:59:57.361+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T09:59:57.361+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T09:59:57.361+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T09:59:57.361+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.1 GiB" free_swap="43.0 GiB" time=2025-10-26T09:59:57.361+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.0 GiB" free="7.5 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T09:59:57.402+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T09:59:57.419+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:57432" time=2025-10-26T09:59:57.428+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:57.444+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T09:59:57.525+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T09:59:57.737+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:26[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:26(22..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:57.759+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:26[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:26(22..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:58.000+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:26[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:26(22..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T09:59:58.000+01:00 level=INFO source=ggml.go:480 msg="offloading 26 repeating layers to GPU" time=2025-10-26T09:59:58.000+01:00 level=INFO source=ggml.go:484 msg="offloading output layer to CPU" time=2025-10-26T09:59:58.000+01:00 level=INFO source=ggml.go:492 msg="offloaded 26/49 layers to GPU" time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.5 GiB" time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.5 GiB" time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="208.0 MiB" time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="176.0 MiB" time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T09:59:58.000+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T09:59:58.000+01:00 level=INFO source=sched.go:482 msg="loaded runners" count=1 time=2025-10-26T09:59:58.000+01:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding" time=2025-10-26T09:59:58.001+01:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model" time=2025-10-26T10:00:11.780+01:00 level=INFO source=server.go:1310 msg="llama runner started in 14.42 seconds" [GIN] 2025/10/26 - 10:00:11 | 200 | 14.7022691s | 127.0.0.1 | POST "/api/generate" [GIN] 2025/10/26 - 10:00:26 | 200 | 0s | 127.0.0.1 | GET "/api/version" [GIN] 2025/10/26 - 10:00:26 | 200 | 529.6µs | 127.0.0.1 | GET "/api/version" [GIN] 2025/10/26 - 10:00:26 | 200 | 7.3573ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/26 - 10:00:26 | 404 | 2.1446ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 10:00:27 | 404 | 3.115ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 10:00:29 | 404 | 2.1104ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 10:01:20 | 200 | 47.8262204s | 127.0.0.1 | POST "/api/chat" [GIN] 2025/10/26 - 10:01:49 | 200 | 0s | 127.0.0.1 | GET "/api/version" ```
GiteaMirror commented 2026-04-12 21:10:11 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

Does it work any better if you set OLLAMA_LLM_LIBRARY=cuda_v12 in the server environment?

<!-- gh-comment-id:3448708987 --> @rick-github commented on GitHub (Oct 26, 2025): Does it work any better if you set `OLLAMA_LLM_LIBRARY=cuda_v12` in the server environment?
GiteaMirror commented 2026-04-12 21:10:12 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

I also can say that ollama 0.9.4 started with the same issue as 0.12.6

<!-- gh-comment-id:3448709145 --> @kiliansinger commented on GitHub (Oct 26, 2025): I also can say that ollama 0.9.4 started with the same issue as 0.12.6
GiteaMirror commented 2026-04-12 21:10:13 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

I am experienced with git bisect if you want I can perform a git bisect.

<!-- gh-comment-id:3448709933 --> @kiliansinger commented on GitHub (Oct 26, 2025): I am experienced with git bisect if you want I can perform a git bisect.
GiteaMirror commented 2026-04-12 21:10:13 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

https://github.com/ollama/ollama/issues/12781#issuecomment-3448708987

<!-- gh-comment-id:3448710526 --> @rick-github commented on GitHub (Oct 26, 2025): https://github.com/ollama/ollama/issues/12781#issuecomment-3448708987
GiteaMirror commented 2026-04-12 21:10:13 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

Same problem on ollama 0.12.6:

>set OLLAMA_LLM_LIBRARY=cuda_v12
>echo %OLLAMA_LLM_LIBRARY%
cuda_v12
>ollama run danielsheep/Qwen3-Coder-30B-A3B-Instruct-1M-Unsloth:UD-IQ3_XXS
Error: 500 Internal Server Error: do load request: Post "http://127.0.0.1:52652/load": read tcp 127.0.0.1:52659->127.0.0.1:52652: wsarecv: An existing connection was forcibly closed by the remote host.
<!-- gh-comment-id:3448712799 --> @kiliansinger commented on GitHub (Oct 26, 2025): Same problem on ollama 0.12.6: ``` >set OLLAMA_LLM_LIBRARY=cuda_v12 >echo %OLLAMA_LLM_LIBRARY% cuda_v12 >ollama run danielsheep/Qwen3-Coder-30B-A3B-Instruct-1M-Unsloth:UD-IQ3_XXS Error: 500 Internal Server Error: do load request: Post "http://127.0.0.1:52652/load": read tcp 127.0.0.1:52659->127.0.0.1:52652: wsarecv: An existing connection was forcibly closed by the remote host. ```
GiteaMirror commented 2026-04-12 21:10:14 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

Does it work any better if you set OLLAMA_LLM_LIBRARY=cuda_v12 in the server environment?

<!-- gh-comment-id:3448715481 --> @rick-github commented on GitHub (Oct 26, 2025): Does it work any better if you set `OLLAMA_LLM_LIBRARY=cuda_v12` in the [**server environment**](https://github.com/ollama/ollama/blob/main/docs/faq.md#setting-environment-variables-on-windows)?
GiteaMirror commented 2026-04-12 21:10:14 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

I did some tests:
And this is the result with OLLAMA_LLM_LIBRARY=cuda_v12
fail,fail,(fail,ok),ok,(fail,fail,ok)
(...) means same login session.

So one can see that repetetive starts can sometimes work.

Then I also did the same in one login session with OLLAMA_LLM_LIBRARY unset.
It failed about10 times and then finally worked in the samesession.
After stopping the model and reloading it it again failed.

<!-- gh-comment-id:3448729700 --> @kiliansinger commented on GitHub (Oct 26, 2025): I did some tests: And this is the result with OLLAMA_LLM_LIBRARY=cuda_v12 fail,fail,(fail,ok),ok,(fail,fail,ok) (...) means same login session. So one can see that repetetive starts can sometimes work. Then I also did the same in one login session with OLLAMA_LLM_LIBRARY unset. It failed about10 times and then finally worked in the samesession. After stopping the model and reloading it it again failed.
GiteaMirror commented 2026-04-12 21:10:14 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

Logs.

<!-- gh-comment-id:3448730582 --> @rick-github commented on GitHub (Oct 26, 2025): Logs.
GiteaMirror commented 2026-04-12 21:10:15 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

I also can be in a login session where none of the starts lead to a success
This is the log for all those starts:

time=2025-10-26T18:43:45.531+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-26T18:43:45.551+01:00 level=INFO source=images.go:522 msg="total blobs: 32"
time=2025-10-26T18:43:45.553+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0"
time=2025-10-26T18:43:45.554+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)"
time=2025-10-26T18:43:45.554+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..."
time=2025-10-26T18:43:48.564+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v13 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB"
time=2025-10-26T18:43:48.564+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB"
[GIN] 2025/10/26 - 18:43:48 | 200 |       535.5µs |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:43:48 | 200 |     34.9583ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:43:48.960+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:43:48.960+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:43:48.960+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:43:48.960+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:43:48.961+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59562"
time=2025-10-26T18:43:48.964+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:43:48.964+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:43:48.964+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:43:48.964+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:43:48.964+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="19.5 GiB" free_swap="49.6 GiB"
time=2025-10-26T18:43:48.964+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:43:48.995+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:43:49.010+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59562"
time=2025-10-26T18:43:49.019+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:43:49.034+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:43:49.189+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:43:49.613+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:43:49.635+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:43:50.035+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59562/load\": read tcp 127.0.0.1:59568->127.0.0.1:59562: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:43:50 | 500 |    1.4163831s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:43:50.064+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:43:52 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:43:52 | 200 |     28.6078ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:43:53.785+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:43:53.785+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:43:53.785+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:43:53.785+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59577"
time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:43:53.790+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:43:53.790+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:43:53.790+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.0 GiB" free_swap="50.1 GiB"
time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:43:53.825+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:43:53.843+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59577"
time=2025-10-26T18:43:53.848+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:43:53.865+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:43:54.028+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:43:54.226+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:43:54.248+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:43:54.632+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59577/load\": read tcp 127.0.0.1:59583->127.0.0.1:59577: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:43:54 | 500 |    1.6690654s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:43:54.665+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:43:56 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:43:56 | 200 |     29.4285ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:43:57.051+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:43:57.051+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:43:57.051+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:43:57.051+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:43:57.052+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59596"
time=2025-10-26T18:43:57.055+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:43:57.055+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:43:57.055+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:43:57.055+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:43:57.055+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.2 GiB" free_swap="50.3 GiB"
time=2025-10-26T18:43:57.055+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:43:57.090+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:43:57.110+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59596"
time=2025-10-26T18:43:57.121+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:43:57.139+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:43:57.298+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:43:57.500+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:43:57.520+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:43:57.918+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59596/load\": read tcp 127.0.0.1:59603->127.0.0.1:59596: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:43:57 | 500 |    1.6539979s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:43:57.947+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:43:59 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:43:59 | 200 |      28.596ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59611"
time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.1 GiB" free_swap="50.1 GiB"
time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:44:00.126+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:44:00.144+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59611"
time=2025-10-26T18:44:00.148+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:00.164+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:44:00.358+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:44:00.571+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:00.592+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:44:00.982+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59611/load\": read tcp 127.0.0.1:59617->127.0.0.1:59611: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:44:00 | 500 |    1.6749859s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:44:01.011+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:44:01 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:44:01 | 200 |     25.7452ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:44:02.525+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:02.525+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:02.525+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:02.525+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:44:02.525+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59632"
time=2025-10-26T18:44:02.529+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:44:02.529+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:02.529+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:02.529+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:02.529+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="19.9 GiB" free_swap="48.8 GiB"
time=2025-10-26T18:44:02.529+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:44:02.562+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:44:02.578+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59632"
time=2025-10-26T18:44:02.585+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:02.600+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:44:02.759+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:44:02.955+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:02.975+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:44:03.343+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59632/load\": read tcp 127.0.0.1:59639->127.0.0.1:59632: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:44:03 | 500 |     1.459115s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:44:03.385+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:44:04 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:44:04 | 200 |     26.3177ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:44:04.964+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:04.964+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:04.964+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:04.964+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59653"
time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:44:04.973+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:04.973+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:04.973+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="19.5 GiB" free_swap="48.3 GiB"
time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:44:05.008+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:44:05.026+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59653"
time=2025-10-26T18:44:05.031+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:05.047+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:44:05.206+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:44:05.401+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:05.422+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:44:05.790+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59653/load\": read tcp 127.0.0.1:59659->127.0.0.1:59653: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:44:05 | 500 |     1.577633s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:44:05.825+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:44:11 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:44:11 | 200 |     26.5905ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:44:11.979+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:11.979+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:11.979+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:11.979+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:44:11.980+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59688"
time=2025-10-26T18:44:11.984+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:44:11.984+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:11.984+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:11.984+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:11.984+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.7 GiB" free_swap="45.6 GiB"
time=2025-10-26T18:44:11.984+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:44:12.029+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:44:12.054+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59688"
time=2025-10-26T18:44:12.061+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:12.078+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:44:12.244+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:44:12.442+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:12.462+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:44:12.845+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:44:12.846+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59688/load\": read tcp 127.0.0.1:65302->127.0.0.1:59688: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:44:12 | 500 |    1.1879497s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:44:12.873+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:44:14 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:44:14 | 200 |     27.8221ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:44:14.608+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:14.608+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:14.608+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:14.608+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:44:14.608+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 52510"
time=2025-10-26T18:44:14.639+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:44:14.639+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:44:14.639+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:44:14.639+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:44:14.639+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.1 GiB" free_swap="44.6 GiB"
time=2025-10-26T18:44:14.639+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:44:14.675+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:44:14.695+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:52510"
time=2025-10-26T18:44:14.705+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:14.719+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:44:14.800+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:44:14.998+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:15.021+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=sched.go:482 msg="loaded runners" count=1
time=2025-10-26T18:44:15.283+01:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=ggml.go:480 msg="offloading 28 repeating layers to GPU"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=ggml.go:484 msg="offloading output layer to CPU"
time=2025-10-26T18:44:15.283+01:00 level=INFO source=ggml.go:492 msg="offloaded 28/49 layers to GPU"
time=2025-10-26T18:44:15.284+01:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model"
time=2025-10-26T18:44:29.811+01:00 level=INFO source=server.go:1310 msg="llama runner started in 15.20 seconds"
[GIN] 2025/10/26 - 18:44:29 | 200 |   15.5545444s |       127.0.0.1 | POST     "/api/generate"
[GIN] 2025/10/26 - 18:44:34 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:44:34 | 200 |     94.4109ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 18:44:34 | 200 |     58.9966ms |       127.0.0.1 | POST     "/api/generate"
[GIN] 2025/10/26 - 18:44:45 | 200 |            0s |       127.0.0.1 | GET      "/api/version"
[GIN] 2025/10/26 - 18:44:45 | 200 |            0s |       127.0.0.1 | GET      "/api/version"
[GIN] 2025/10/26 - 18:44:45 | 200 |      4.8616ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/26 - 18:44:45 | 404 |      1.5779ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 18:44:46 | 404 |      1.5693ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 18:44:48 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:44:48 | 200 |     32.0634ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 18:44:48 | 200 |      54.809ms |       127.0.0.1 | POST     "/api/generate"
[GIN] 2025/10/26 - 18:46:05 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:46:05 | 200 |     29.7927ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 18:46:05 | 200 |     56.3293ms |       127.0.0.1 | POST     "/api/generate"
[GIN] 2025/10/26 - 18:46:22 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:46:22 | 200 |      2.1506ms |       127.0.0.1 | POST     "/api/generate"
ggml_backend_cuda_device_get_memory utilizing NVML memory reporting free: 329535488 total: 8585740288
[GIN] 2025/10/26 - 18:46:27 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:46:27 | 200 |     29.6879ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:46:28.105+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:46:28.105+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:46:28.105+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:46:28.105+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:46:28.106+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 53521"
time=2025-10-26T18:46:28.109+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:46:28.109+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:46:28.109+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:46:28.109+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:46:28.109+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.3 GiB" free_swap="41.4 GiB"
time=2025-10-26T18:46:28.109+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:46:28.147+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:46:28.165+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:53521"
time=2025-10-26T18:46:28.177+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:46:28.194+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:46:28.277+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:46:28.543+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:46:28.564+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:46:28.986+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:53521/load\": read tcp 127.0.0.1:53528->127.0.0.1:53521: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:46:28 | 500 |    1.1472558s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:46:29.018+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:46:37 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:46:37 | 200 |     31.1901ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:46:37.471+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:46:37.471+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:46:37.471+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:46:37.471+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:46:37.472+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 53543"
time=2025-10-26T18:46:37.477+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:46:37.477+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:46:37.477+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:46:37.477+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:46:37.477+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.3 GiB" free_swap="41.4 GiB"
time=2025-10-26T18:46:37.477+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:46:37.518+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:46:37.535+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:53543"
time=2025-10-26T18:46:37.543+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:46:37.559+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:46:37.642+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:46:37.861+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:46:37.883+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:46:38.236+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:53543/load\": read tcp 127.0.0.1:53550->127.0.0.1:53543: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:46:38 | 500 |    1.0316261s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:46:38.267+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:22 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:22 | 200 |     31.2289ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:22.615+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:22.615+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:22.615+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:22.615+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:22.616+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 58379"
time=2025-10-26T18:51:22.619+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:22.619+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:22.619+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:22.619+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:22.619+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.6 GiB" free_swap="40.1 GiB"
time=2025-10-26T18:51:22.620+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:22.657+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:22.675+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:58379"
time=2025-10-26T18:51:22.685+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:22.704+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:22.788+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:23.009+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:23.030+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:23.445+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:58379/load\": read tcp 127.0.0.1:58386->127.0.0.1:58379: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:23 | 500 |    1.1153101s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:23.472+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:35 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:35 | 200 |     48.9568ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:35.714+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:35.714+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:35.714+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:35.714+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:35.715+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49880"
time=2025-10-26T18:51:35.720+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:35.720+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:35.720+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:35.720+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:35.720+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.5 GiB" free_swap="40.0 GiB"
time=2025-10-26T18:51:35.720+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:35.766+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:35.783+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49880"
time=2025-10-26T18:51:35.790+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:35.807+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:35.892+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:36.119+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:36.143+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:36.534+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49880/load\": read tcp 127.0.0.1:49887->127.0.0.1:49880: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:36 | 500 |    1.2591479s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:36.562+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:37 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:37 | 200 |     54.7868ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:38.193+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:38.193+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:38.193+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:38.193+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:38.194+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49899"
time=2025-10-26T18:51:38.199+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:38.199+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:38.199+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:38.199+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:38.199+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:38.199+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:38.244+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:38.261+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49899"
time=2025-10-26T18:51:38.270+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:38.287+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:38.371+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:38.593+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:38.617+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:38.991+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49899/load\": read tcp 127.0.0.1:49906->127.0.0.1:49899: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:38 | 500 |    1.2019763s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:39.020+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:39 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:39 | 200 |     50.8257ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:40.422+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:40.422+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:40.422+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:40.422+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:40.433+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49919"
time=2025-10-26T18:51:40.438+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:40.438+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:40.438+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:40.438+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:40.438+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:40.438+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:40.483+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:40.500+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49919"
time=2025-10-26T18:51:40.505+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:40.522+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:40.606+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:40.822+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:40.846+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:41.236+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49919/load\": read tcp 127.0.0.1:49926->127.0.0.1:49919: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:41 | 500 |    1.2277169s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:41.263+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:41 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:41 | 200 |     48.5271ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:42.293+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:42.293+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:42.293+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:42.293+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:42.294+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49938"
time=2025-10-26T18:51:42.298+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:42.298+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:42.298+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:42.298+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:42.298+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:42.298+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:42.343+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:42.360+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49938"
time=2025-10-26T18:51:42.367+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:42.384+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:42.467+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:42.683+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:42.706+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:43.031+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49938/load\": read tcp 127.0.0.1:49945->127.0.0.1:49938: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:43 | 500 |    1.1507266s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:43.059+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:43 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:43 | 200 |     50.5493ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:44.158+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:44.158+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:44.158+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:44.158+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:44.160+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49957"
time=2025-10-26T18:51:44.165+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:44.165+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:44.165+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:44.165+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:44.165+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:44.165+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:44.210+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:44.228+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49957"
time=2025-10-26T18:51:44.233+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:44.250+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:44.334+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:44.546+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:44.569+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:44.889+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49957/load\": read tcp 127.0.0.1:49964->127.0.0.1:49957: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:44 | 500 |    1.1461117s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:44.916+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:45 | 200 |       518.8µs |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:45 | 200 |     46.8272ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:45.980+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:45.980+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:45.980+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:45.980+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:45.982+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49981"
time=2025-10-26T18:51:45.987+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:45.987+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:45.987+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:45.987+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:45.987+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:45.987+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:46.030+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:46.047+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49981"
time=2025-10-26T18:51:46.056+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:46.074+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:46.159+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:46.374+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:46.397+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:46.723+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49981/load\": read tcp 127.0.0.1:49988->127.0.0.1:49981: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:46 | 500 |    1.1704949s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:46.752+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:47 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:47 | 200 |     47.2438ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:47.751+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:47.751+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:47.751+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:47.751+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:47.752+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50161"
time=2025-10-26T18:51:47.757+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:47.758+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:47.758+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:47.758+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:47.758+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:47.758+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:47.800+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:47.817+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50161"
time=2025-10-26T18:51:47.826+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:47.843+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:47.926+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:48.145+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:48.167+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:48.490+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50161/load\": read tcp 127.0.0.1:50369->127.0.0.1:50161: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:48 | 500 |    1.1489052s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:48.522+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:49 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:49 | 200 |     50.0316ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:49.629+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:49.629+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:49.629+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:49.629+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:49.631+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50381"
time=2025-10-26T18:51:49.635+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:49.635+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:49.635+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:49.635+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:49.635+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.1 GiB"
time=2025-10-26T18:51:49.635+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:49.681+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:49.700+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50381"
time=2025-10-26T18:51:49.704+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:49.721+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:49.803+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:50.021+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:50.045+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:50.368+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50381/load\": read tcp 127.0.0.1:50388->127.0.0.1:50381: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:50 | 500 |    1.1616057s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:50.395+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:50 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:51 | 200 |     51.0879ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:51.477+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:51.477+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:51.477+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:51.477+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:51.478+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50399"
time=2025-10-26T18:51:51.484+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:51.484+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:51.484+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:51.484+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:51.484+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:51.484+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:51.533+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:51.551+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50399"
time=2025-10-26T18:51:51.563+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:51.581+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:51.664+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:51.886+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:51.910+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:52.295+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50399/load\": read tcp 127.0.0.1:50407->127.0.0.1:50399: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:52 | 500 |     1.229501s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:52.322+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:52 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:52 | 200 |     47.6754ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:53.299+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:53.300+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:53.300+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:53.300+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:53.301+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50420"
time=2025-10-26T18:51:53.306+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:53.306+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:53.306+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:53.306+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:53.306+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:53.306+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:53.351+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:53.368+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50420"
time=2025-10-26T18:51:53.375+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:53.392+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:53.479+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:53.691+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:53.714+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:54.035+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50420/load\": read tcp 127.0.0.1:50427->127.0.0.1:50420: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:54 | 500 |    1.1545916s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:54.062+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:54 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:54 | 200 |     53.8563ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:55.089+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:55.089+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:55.089+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:55.089+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:55.091+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50442"
time=2025-10-26T18:51:55.094+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:55.094+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:55.094+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:55.094+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:55.094+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:55.094+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:55.138+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:55.156+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50442"
time=2025-10-26T18:51:55.165+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:55.181+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:55.265+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:55.482+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:55.504+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:55.825+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50442/load\": read tcp 127.0.0.1:50449->127.0.0.1:50442: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:55 | 500 |    1.1524826s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:55.853+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:56 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:56 | 200 |     49.6194ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:56.863+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:56.863+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:56.863+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:56.863+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:56.864+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50461"
time=2025-10-26T18:51:56.868+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:56.869+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:56.869+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:56.869+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:56.869+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:56.869+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:56.912+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:56.930+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50461"
time=2025-10-26T18:51:56.936+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:56.953+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:57.035+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:57.247+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:57.270+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:57.592+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50461/load\": read tcp 127.0.0.1:50468->127.0.0.1:50461: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:57 | 500 |    1.1471221s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:57.619+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:51:58 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:51:58 | 200 |      48.628ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:51:58.651+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:58.651+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:58.651+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:58.651+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:51:58.653+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50480"
time=2025-10-26T18:51:58.658+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:51:58.658+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:51:58.658+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:51:58.658+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:51:58.658+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:51:58.658+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:51:58.705+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:51:58.723+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50480"
time=2025-10-26T18:51:58.726+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:58.744+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:51:58.827+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:51:59.039+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:51:59.063+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:51:59.387+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50480/load\": read tcp 127.0.0.1:50487->127.0.0.1:50480: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:51:59 | 500 |     1.144958s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:51:59.414+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:03 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:03 | 200 |      3.8578ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/26 - 18:52:06 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:06 | 200 |      1.6398ms |       127.0.0.1 | GET      "/api/ps"
[GIN] 2025/10/26 - 18:52:10 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:10 | 200 |     50.4214ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:11.255+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:11.255+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:11.255+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:11.255+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:11.257+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50546"
time=2025-10-26T18:52:11.261+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:11.261+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:11.261+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:11.261+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:11.261+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:11.261+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:11.303+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:11.320+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50546"
time=2025-10-26T18:52:11.329+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:11.345+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:11.429+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:11.657+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:11.680+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:12.071+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50546/load\": read tcp 127.0.0.1:50553->127.0.0.1:50546: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:12 | 500 |     1.222992s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:12.098+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:13 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:13 | 200 |     49.2099ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:13.478+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:13.478+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:13.478+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:13.478+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:13.480+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50565"
time=2025-10-26T18:52:13.484+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:13.484+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:13.484+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:13.484+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:13.484+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:13.484+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:13.528+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:13.546+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50565"
time=2025-10-26T18:52:13.553+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:13.569+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:13.653+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:13.865+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:13.887+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:14.278+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50565/load\": read tcp 127.0.0.1:50572->127.0.0.1:50565: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:14 | 500 |    1.2058043s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:14.305+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:14 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:14 | 200 |     50.9243ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:15.348+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:15.348+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:15.348+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:15.348+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:15.349+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50587"
time=2025-10-26T18:52:15.353+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:15.353+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:15.353+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:15.353+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:15.353+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:15.354+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:15.399+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:15.417+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50587"
time=2025-10-26T18:52:15.423+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:15.439+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:15.521+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:15.732+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:15.754+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:16.078+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50587/load\": read tcp 127.0.0.1:50594->127.0.0.1:50587: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:16 | 500 |    1.1362903s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:16.106+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:16 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:16 | 200 |     50.8232ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:17.136+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:17.136+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:17.136+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:17.136+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:17.138+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50606"
time=2025-10-26T18:52:17.143+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:17.143+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:17.143+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:17.143+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:17.143+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:17.143+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:17.188+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:17.205+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50606"
time=2025-10-26T18:52:17.211+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:17.227+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:17.311+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:17.521+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:17.543+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:17.865+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50606/load\": read tcp 127.0.0.1:50613->127.0.0.1:50606: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:17 | 500 |    1.1486882s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:17.893+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:18 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:18 | 200 |     49.0537ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:18.840+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:18.840+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:18.840+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:18.840+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:18.841+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50625"
time=2025-10-26T18:52:18.845+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:18.845+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:18.845+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:18.845+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:18.845+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:18.845+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:18.887+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:18.905+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50625"
time=2025-10-26T18:52:18.915+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:18.932+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:19.013+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:19.222+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:19.244+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:19.578+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50625/load\": read tcp 127.0.0.1:50632->127.0.0.1:50625: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:19 | 500 |     1.156624s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:19.605+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:20 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:20 | 200 |     48.0629ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:20.697+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:20.697+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:20.697+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:20.697+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:20.699+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50644"
time=2025-10-26T18:52:20.703+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:20.703+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:20.703+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:20.703+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:20.703+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:20.703+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:20.750+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:20.767+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50644"
time=2025-10-26T18:52:20.773+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:20.790+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:20.872+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:21.088+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:21.110+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:21.496+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50644/load\": read tcp 127.0.0.1:50651->127.0.0.1:50644: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:21 | 500 |    1.2150414s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:21.523+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:21 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:21 | 200 |     46.2977ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:22.411+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:22.411+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:22.411+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:22.411+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:22.413+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50663"
time=2025-10-26T18:52:22.416+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:22.416+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:22.416+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:22.416+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:22.416+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:22.416+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:22.460+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:22.477+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50663"
time=2025-10-26T18:52:22.485+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:22.501+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:22.585+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:22.797+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:22.819+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:23.143+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50663/load\": read tcp 127.0.0.1:50670->127.0.0.1:50663: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:23 | 500 |    1.1476022s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:23.170+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:23 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:23 | 200 |     50.5666ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:24.058+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:24.058+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:24.058+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:24.058+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:24.059+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50682"
time=2025-10-26T18:52:24.064+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:24.064+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:24.064+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:24.064+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:24.064+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:24.064+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:24.110+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:24.127+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50682"
time=2025-10-26T18:52:24.133+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:24.149+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:24.231+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:24.444+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:24.466+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:24.790+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50682/load\": read tcp 127.0.0.1:50689->127.0.0.1:50682: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:24 | 500 |    1.1418636s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:24.818+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:25 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:25 | 200 |     47.6571ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:25.682+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:25.682+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:25.682+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:25.682+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:25.683+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50704"
time=2025-10-26T18:52:25.689+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:25.689+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:25.689+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:25.689+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:25.689+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:25.690+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:25.734+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:25.751+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50704"
time=2025-10-26T18:52:25.758+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:25.776+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:25.858+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:26.071+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:26.091+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:26.415+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:26.416+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50704/load\": read tcp 127.0.0.1:50711->127.0.0.1:50704: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:26 | 500 |    1.1388094s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:26.443+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:26 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:26 | 200 |     48.6038ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:27.262+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:27.262+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:27.262+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:27.262+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:27.272+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50723"
time=2025-10-26T18:52:27.277+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:27.277+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:27.277+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:27.277+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:27.278+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:27.278+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:27.322+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:27.339+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50723"
time=2025-10-26T18:52:27.346+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:27.363+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:27.449+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:27.661+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:27.684+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:28.011+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50723/load\": read tcp 127.0.0.1:50730->127.0.0.1:50723: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:28 | 500 |    1.1726113s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:28.045+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:28 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:28 | 200 |     50.6107ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:28.880+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:28.880+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:28.880+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:28.880+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:28.882+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50747"
time=2025-10-26T18:52:28.887+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:28.887+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:28.887+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:28.887+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:28.887+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:28.887+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:28.932+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:28.949+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50747"
time=2025-10-26T18:52:28.956+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:28.972+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:29.059+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:29.267+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:29.289+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:29.617+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50747/load\": read tcp 127.0.0.1:50754->127.0.0.1:50747: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:29 | 500 |    1.1424097s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:29.646+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 18:52:30 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 18:52:30 | 200 |     57.7387ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T18:52:30.505+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:30.505+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:30.505+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:30.505+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T18:52:30.506+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50766"
time=2025-10-26T18:52:30.510+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T18:52:30.510+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T18:52:30.510+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T18:52:30.510+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T18:52:30.510+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB"
time=2025-10-26T18:52:30.510+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T18:52:30.557+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T18:52:30.574+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50766"
time=2025-10-26T18:52:30.579+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:30.595+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll
time=2025-10-26T18:52:30.679+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T18:52:30.888+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T18:52:30.909+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T18:52:31.232+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50766/load\": read tcp 127.0.0.1:50773->127.0.0.1:50766: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 18:52:31 | 500 |    1.1375198s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T18:52:31.260+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
<!-- gh-comment-id:3448731886 --> @kiliansinger commented on GitHub (Oct 26, 2025): I also can be in a login session where none of the starts lead to a success This is the log for all those starts: ``` time=2025-10-26T18:43:45.531+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-26T18:43:45.551+01:00 level=INFO source=images.go:522 msg="total blobs: 32" time=2025-10-26T18:43:45.553+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0" time=2025-10-26T18:43:45.554+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)" time=2025-10-26T18:43:45.554+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..." time=2025-10-26T18:43:48.564+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v13 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB" time=2025-10-26T18:43:48.564+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB" [GIN] 2025/10/26 - 18:43:48 | 200 | 535.5µs | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:43:48 | 200 | 34.9583ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:43:48.960+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:43:48.960+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:43:48.960+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:43:48.960+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:43:48.961+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59562" time=2025-10-26T18:43:48.964+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:43:48.964+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:43:48.964+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:43:48.964+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:43:48.964+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="19.5 GiB" free_swap="49.6 GiB" time=2025-10-26T18:43:48.964+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:43:48.995+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:43:49.010+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59562" time=2025-10-26T18:43:49.019+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:43:49.034+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:43:49.189+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:43:49.613+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:43:49.635+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:43:50.035+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:43:50.035+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59562/load\": read tcp 127.0.0.1:59568->127.0.0.1:59562: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:43:50 | 500 | 1.4163831s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:43:50.064+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:43:52 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:43:52 | 200 | 28.6078ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:43:53.785+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:43:53.785+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:43:53.785+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:43:53.785+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59577" time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:43:53.790+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:43:53.790+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:43:53.790+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.0 GiB" free_swap="50.1 GiB" time=2025-10-26T18:43:53.790+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:43:53.825+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:43:53.843+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59577" time=2025-10-26T18:43:53.848+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:43:53.865+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:43:54.028+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:43:54.226+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:43:54.248+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:43:54.632+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:43:54.632+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59577/load\": read tcp 127.0.0.1:59583->127.0.0.1:59577: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:43:54 | 500 | 1.6690654s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:43:54.665+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:43:56 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:43:56 | 200 | 29.4285ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:43:57.051+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:43:57.051+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:43:57.051+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:43:57.051+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:43:57.052+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59596" time=2025-10-26T18:43:57.055+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:43:57.055+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:43:57.055+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:43:57.055+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:43:57.055+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.2 GiB" free_swap="50.3 GiB" time=2025-10-26T18:43:57.055+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:43:57.090+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:43:57.110+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59596" time=2025-10-26T18:43:57.121+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:43:57.139+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:43:57.298+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:43:57.500+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:43:57.520+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:43:57.918+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:43:57.918+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59596/load\": read tcp 127.0.0.1:59603->127.0.0.1:59596: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:43:57 | 500 | 1.6539979s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:43:57.947+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:43:59 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:43:59 | 200 | 28.596ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59611" time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:00.088+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="20.1 GiB" free_swap="50.1 GiB" time=2025-10-26T18:44:00.088+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:44:00.126+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:44:00.144+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59611" time=2025-10-26T18:44:00.148+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:00.164+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:44:00.358+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:44:00.571+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:00.592+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:44:00.982+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:44:00.982+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59611/load\": read tcp 127.0.0.1:59617->127.0.0.1:59611: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:44:00 | 500 | 1.6749859s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:44:01.011+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:44:01 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:44:01 | 200 | 25.7452ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:44:02.525+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:02.525+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:02.525+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:02.525+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:44:02.525+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59632" time=2025-10-26T18:44:02.529+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:44:02.529+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:02.529+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:02.529+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:02.529+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="19.9 GiB" free_swap="48.8 GiB" time=2025-10-26T18:44:02.529+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:44:02.562+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:44:02.578+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59632" time=2025-10-26T18:44:02.585+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:02.600+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:44:02.759+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:44:02.955+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:02.975+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:44:03.343+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:44:03.343+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59632/load\": read tcp 127.0.0.1:59639->127.0.0.1:59632: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:44:03 | 500 | 1.459115s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:44:03.385+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:44:04 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:44:04 | 200 | 26.3177ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:44:04.964+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:04.964+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:04.964+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:04.964+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59653" time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:44:04.973+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:04.973+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:04.973+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="19.5 GiB" free_swap="48.3 GiB" time=2025-10-26T18:44:04.973+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:44:05.008+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:44:05.026+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59653" time=2025-10-26T18:44:05.031+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:05.047+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:44:05.206+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:44:05.401+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:05.422+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:44:05.790+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:44:05.790+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59653/load\": read tcp 127.0.0.1:59659->127.0.0.1:59653: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:44:05 | 500 | 1.577633s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:44:05.825+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:44:11 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:44:11 | 200 | 26.5905ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:44:11.979+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:11.979+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:11.979+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:11.979+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:44:11.980+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59688" time=2025-10-26T18:44:11.984+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:44:11.984+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:11.984+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:11.984+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:11.984+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.7 GiB" free_swap="45.6 GiB" time=2025-10-26T18:44:11.984+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:44:12.029+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:44:12.054+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59688" time=2025-10-26T18:44:12.061+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:12.078+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:44:12.244+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:44:12.442+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:12.462+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:44:12.845+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:44:12.846+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:44:12.846+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59688/load\": read tcp 127.0.0.1:65302->127.0.0.1:59688: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:44:12 | 500 | 1.1879497s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:44:12.873+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:44:14 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:44:14 | 200 | 27.8221ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:44:14.608+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:14.608+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:14.608+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:14.608+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:44:14.608+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 52510" time=2025-10-26T18:44:14.639+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:44:14.639+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:44:14.639+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:44:14.639+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:44:14.639+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.1 GiB" free_swap="44.6 GiB" time=2025-10-26T18:44:14.639+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:44:14.675+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:44:14.695+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:52510" time=2025-10-26T18:44:14.705+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:14.719+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:44:14.800+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:44:14.998+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:15.021+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:15.283+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:44:15.283+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:44:15.283+01:00 level=INFO source=sched.go:482 msg="loaded runners" count=1 time=2025-10-26T18:44:15.283+01:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding" time=2025-10-26T18:44:15.283+01:00 level=INFO source=ggml.go:480 msg="offloading 28 repeating layers to GPU" time=2025-10-26T18:44:15.283+01:00 level=INFO source=ggml.go:484 msg="offloading output layer to CPU" time=2025-10-26T18:44:15.283+01:00 level=INFO source=ggml.go:492 msg="offloaded 28/49 layers to GPU" time=2025-10-26T18:44:15.284+01:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model" time=2025-10-26T18:44:29.811+01:00 level=INFO source=server.go:1310 msg="llama runner started in 15.20 seconds" [GIN] 2025/10/26 - 18:44:29 | 200 | 15.5545444s | 127.0.0.1 | POST "/api/generate" [GIN] 2025/10/26 - 18:44:34 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:44:34 | 200 | 94.4109ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 18:44:34 | 200 | 58.9966ms | 127.0.0.1 | POST "/api/generate" [GIN] 2025/10/26 - 18:44:45 | 200 | 0s | 127.0.0.1 | GET "/api/version" [GIN] 2025/10/26 - 18:44:45 | 200 | 0s | 127.0.0.1 | GET "/api/version" [GIN] 2025/10/26 - 18:44:45 | 200 | 4.8616ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/26 - 18:44:45 | 404 | 1.5779ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 18:44:46 | 404 | 1.5693ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 18:44:48 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:44:48 | 200 | 32.0634ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 18:44:48 | 200 | 54.809ms | 127.0.0.1 | POST "/api/generate" [GIN] 2025/10/26 - 18:46:05 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:46:05 | 200 | 29.7927ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 18:46:05 | 200 | 56.3293ms | 127.0.0.1 | POST "/api/generate" [GIN] 2025/10/26 - 18:46:22 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:46:22 | 200 | 2.1506ms | 127.0.0.1 | POST "/api/generate" ggml_backend_cuda_device_get_memory utilizing NVML memory reporting free: 329535488 total: 8585740288 [GIN] 2025/10/26 - 18:46:27 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:46:27 | 200 | 29.6879ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:46:28.105+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:46:28.105+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:46:28.105+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:46:28.105+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:46:28.106+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 53521" time=2025-10-26T18:46:28.109+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:46:28.109+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:46:28.109+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:46:28.109+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:46:28.109+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.3 GiB" free_swap="41.4 GiB" time=2025-10-26T18:46:28.109+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:46:28.147+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:46:28.165+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:53521" time=2025-10-26T18:46:28.177+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:46:28.194+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:46:28.277+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:46:28.543+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:46:28.564+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:46:28.986+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:46:28.986+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:53521/load\": read tcp 127.0.0.1:53528->127.0.0.1:53521: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:46:28 | 500 | 1.1472558s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:46:29.018+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:46:37 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:46:37 | 200 | 31.1901ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:46:37.471+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:46:37.471+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:46:37.471+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:46:37.471+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:46:37.472+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 53543" time=2025-10-26T18:46:37.477+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:46:37.477+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:46:37.477+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:46:37.477+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:46:37.477+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.3 GiB" free_swap="41.4 GiB" time=2025-10-26T18:46:37.477+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:46:37.518+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:46:37.535+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:53543" time=2025-10-26T18:46:37.543+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:46:37.559+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:46:37.642+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:46:37.861+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:46:37.883+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:46:38.236+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:46:38.236+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:53543/load\": read tcp 127.0.0.1:53550->127.0.0.1:53543: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:46:38 | 500 | 1.0316261s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:46:38.267+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:22 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:22 | 200 | 31.2289ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:22.615+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:22.615+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:22.615+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:22.615+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:22.616+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 58379" time=2025-10-26T18:51:22.619+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:22.619+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:22.619+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:22.619+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:22.619+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.6 GiB" free_swap="40.1 GiB" time=2025-10-26T18:51:22.620+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:22.657+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:22.675+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:58379" time=2025-10-26T18:51:22.685+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:22.704+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:22.788+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:23.009+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:23.030+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:23.445+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:23.445+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:58379/load\": read tcp 127.0.0.1:58386->127.0.0.1:58379: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:23 | 500 | 1.1153101s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:23.472+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:35 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:35 | 200 | 48.9568ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:35.714+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:35.714+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:35.714+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:35.714+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:35.715+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49880" time=2025-10-26T18:51:35.720+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:35.720+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:35.720+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:35.720+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:35.720+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.5 GiB" free_swap="40.0 GiB" time=2025-10-26T18:51:35.720+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:35.766+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:35.783+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49880" time=2025-10-26T18:51:35.790+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:35.807+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:35.892+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:36.119+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:36.143+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:36.534+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:36.534+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49880/load\": read tcp 127.0.0.1:49887->127.0.0.1:49880: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:36 | 500 | 1.2591479s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:36.562+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:37 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:37 | 200 | 54.7868ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:38.193+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:38.193+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:38.193+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:38.193+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:38.194+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49899" time=2025-10-26T18:51:38.199+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:38.199+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:38.199+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:38.199+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:38.199+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:38.199+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:38.244+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:38.261+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49899" time=2025-10-26T18:51:38.270+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:38.287+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:38.371+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:38.593+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:38.617+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:38.991+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:38.991+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49899/load\": read tcp 127.0.0.1:49906->127.0.0.1:49899: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:38 | 500 | 1.2019763s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:39.020+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:39 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:39 | 200 | 50.8257ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:40.422+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:40.422+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:40.422+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:40.422+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:40.433+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49919" time=2025-10-26T18:51:40.438+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:40.438+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:40.438+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:40.438+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:40.438+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:40.438+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:40.483+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:40.500+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49919" time=2025-10-26T18:51:40.505+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:40.522+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:40.606+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:40.822+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:40.846+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:41.236+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:41.236+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49919/load\": read tcp 127.0.0.1:49926->127.0.0.1:49919: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:41 | 500 | 1.2277169s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:41.263+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:41 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:41 | 200 | 48.5271ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:42.293+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:42.293+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:42.293+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:42.293+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:42.294+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49938" time=2025-10-26T18:51:42.298+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:42.298+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:42.298+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:42.298+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:42.298+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:42.298+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:42.343+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:42.360+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49938" time=2025-10-26T18:51:42.367+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:42.384+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:42.467+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:42.683+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:42.706+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:43.031+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:43.031+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49938/load\": read tcp 127.0.0.1:49945->127.0.0.1:49938: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:43 | 500 | 1.1507266s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:43.059+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:43 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:43 | 200 | 50.5493ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:44.158+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:44.158+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:44.158+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:44.158+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:44.160+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49957" time=2025-10-26T18:51:44.165+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:44.165+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:44.165+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:44.165+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:44.165+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:44.165+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:44.210+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:44.228+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49957" time=2025-10-26T18:51:44.233+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:44.250+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:44.334+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:44.546+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:44.569+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:44.889+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:44.889+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49957/load\": read tcp 127.0.0.1:49964->127.0.0.1:49957: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:44 | 500 | 1.1461117s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:44.916+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:45 | 200 | 518.8µs | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:45 | 200 | 46.8272ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:45.980+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:45.980+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:45.980+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:45.980+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:45.982+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49981" time=2025-10-26T18:51:45.987+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:45.987+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:45.987+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:45.987+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:45.987+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:45.987+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:46.030+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:46.047+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49981" time=2025-10-26T18:51:46.056+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:46.074+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:46.159+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:46.374+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:46.397+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:46.723+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:46.723+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:49981/load\": read tcp 127.0.0.1:49988->127.0.0.1:49981: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:46 | 500 | 1.1704949s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:46.752+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:47 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:47 | 200 | 47.2438ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:47.751+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:47.751+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:47.751+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:47.751+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:47.752+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50161" time=2025-10-26T18:51:47.757+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:47.758+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:47.758+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:47.758+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:47.758+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:47.758+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:47.800+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:47.817+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50161" time=2025-10-26T18:51:47.826+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:47.843+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:47.926+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:48.145+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:48.167+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:48.490+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:48.490+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50161/load\": read tcp 127.0.0.1:50369->127.0.0.1:50161: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:48 | 500 | 1.1489052s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:48.522+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:49 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:49 | 200 | 50.0316ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:49.629+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:49.629+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:49.629+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:49.629+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:49.631+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50381" time=2025-10-26T18:51:49.635+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:49.635+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:49.635+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:49.635+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:49.635+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.1 GiB" time=2025-10-26T18:51:49.635+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:49.681+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:49.700+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50381" time=2025-10-26T18:51:49.704+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:49.721+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:49.803+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:50.021+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:50.045+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:50.368+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:50.368+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50381/load\": read tcp 127.0.0.1:50388->127.0.0.1:50381: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:50 | 500 | 1.1616057s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:50.395+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:50 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:51 | 200 | 51.0879ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:51.477+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:51.477+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:51.477+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:51.477+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:51.478+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50399" time=2025-10-26T18:51:51.484+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:51.484+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:51.484+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:51.484+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:51.484+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:51.484+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:51.533+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:51.551+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50399" time=2025-10-26T18:51:51.563+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:51.581+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:51.664+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:51.886+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:51.910+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:52.294+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:52.295+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50399/load\": read tcp 127.0.0.1:50407->127.0.0.1:50399: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:52 | 500 | 1.229501s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:52.322+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:52 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:52 | 200 | 47.6754ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:53.299+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:53.300+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:53.300+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:53.300+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:53.301+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50420" time=2025-10-26T18:51:53.306+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:53.306+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:53.306+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:53.306+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:53.306+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:53.306+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:53.351+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:53.368+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50420" time=2025-10-26T18:51:53.375+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:53.392+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:53.479+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:53.691+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:53.714+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:54.035+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:54.035+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50420/load\": read tcp 127.0.0.1:50427->127.0.0.1:50420: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:54 | 500 | 1.1545916s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:54.062+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:54 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:54 | 200 | 53.8563ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:55.089+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:55.089+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:55.089+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:55.089+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:55.091+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50442" time=2025-10-26T18:51:55.094+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:55.094+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:55.094+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:55.094+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:55.094+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:55.094+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:55.138+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:55.156+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50442" time=2025-10-26T18:51:55.165+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:55.181+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:55.265+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:55.482+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:55.504+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:55.825+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:55.825+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50442/load\": read tcp 127.0.0.1:50449->127.0.0.1:50442: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:55 | 500 | 1.1524826s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:55.853+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:56 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:56 | 200 | 49.6194ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:56.863+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:56.863+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:56.863+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:56.863+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:56.864+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50461" time=2025-10-26T18:51:56.868+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:56.869+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:56.869+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:56.869+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:56.869+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:56.869+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:56.912+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:56.930+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50461" time=2025-10-26T18:51:56.936+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:56.953+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:57.035+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:57.247+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:57.270+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:57.592+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:57.592+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50461/load\": read tcp 127.0.0.1:50468->127.0.0.1:50461: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:57 | 500 | 1.1471221s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:57.619+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:51:58 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:51:58 | 200 | 48.628ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:51:58.651+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:58.651+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:58.651+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:58.651+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:51:58.653+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50480" time=2025-10-26T18:51:58.658+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:51:58.658+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:51:58.658+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:51:58.658+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:51:58.658+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:51:58.658+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:51:58.705+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:51:58.723+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50480" time=2025-10-26T18:51:58.726+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:58.744+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:51:58.827+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:51:59.039+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:51:59.063+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:51:59.387+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:51:59.387+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50480/load\": read tcp 127.0.0.1:50487->127.0.0.1:50480: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:51:59 | 500 | 1.144958s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:51:59.414+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:03 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:03 | 200 | 3.8578ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/26 - 18:52:06 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:06 | 200 | 1.6398ms | 127.0.0.1 | GET "/api/ps" [GIN] 2025/10/26 - 18:52:10 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:10 | 200 | 50.4214ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:11.255+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:11.255+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:11.255+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:11.255+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:11.257+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50546" time=2025-10-26T18:52:11.261+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:11.261+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:11.261+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:11.261+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:11.261+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.7 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:11.261+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:11.303+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:11.320+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50546" time=2025-10-26T18:52:11.329+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:11.345+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:11.429+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:11.657+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:11.680+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:12.071+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:12.071+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50546/load\": read tcp 127.0.0.1:50553->127.0.0.1:50546: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:12 | 500 | 1.222992s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:12.098+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:13 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:13 | 200 | 49.2099ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:13.478+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:13.478+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:13.478+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:13.478+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:13.480+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50565" time=2025-10-26T18:52:13.484+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:13.484+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:13.484+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:13.484+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:13.484+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:13.484+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:13.528+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:13.546+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50565" time=2025-10-26T18:52:13.553+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:13.569+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:13.653+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:13.865+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:13.887+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:14.278+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:14.278+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50565/load\": read tcp 127.0.0.1:50572->127.0.0.1:50565: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:14 | 500 | 1.2058043s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:14.305+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:14 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:14 | 200 | 50.9243ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:15.348+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:15.348+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:15.348+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:15.348+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:15.349+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50587" time=2025-10-26T18:52:15.353+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:15.353+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:15.353+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:15.353+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:15.353+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:15.354+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:15.399+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:15.417+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50587" time=2025-10-26T18:52:15.423+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:15.439+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:15.521+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:15.732+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:15.754+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:16.078+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:16.078+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50587/load\": read tcp 127.0.0.1:50594->127.0.0.1:50587: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:16 | 500 | 1.1362903s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:16.106+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:16 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:16 | 200 | 50.8232ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:17.136+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:17.136+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:17.136+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:17.136+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:17.138+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50606" time=2025-10-26T18:52:17.143+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:17.143+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:17.143+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:17.143+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:17.143+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:17.143+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:17.188+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:17.205+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50606" time=2025-10-26T18:52:17.211+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:17.227+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:17.311+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:17.521+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:17.543+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:17.865+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:17.865+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50606/load\": read tcp 127.0.0.1:50613->127.0.0.1:50606: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:17 | 500 | 1.1486882s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:17.893+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:18 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:18 | 200 | 49.0537ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:18.840+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:18.840+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:18.840+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:18.840+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:18.841+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50625" time=2025-10-26T18:52:18.845+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:18.845+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:18.845+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:18.845+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:18.845+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:18.845+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:18.887+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:18.905+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50625" time=2025-10-26T18:52:18.915+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:18.932+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:19.013+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:19.222+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:19.244+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:19.577+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:19.578+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50625/load\": read tcp 127.0.0.1:50632->127.0.0.1:50625: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:19 | 500 | 1.156624s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:19.605+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:20 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:20 | 200 | 48.0629ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:20.697+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:20.697+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:20.697+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:20.697+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:20.699+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50644" time=2025-10-26T18:52:20.703+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:20.703+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:20.703+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:20.703+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:20.703+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:20.703+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:20.750+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:20.767+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50644" time=2025-10-26T18:52:20.773+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:20.790+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:20.872+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:21.088+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:21.110+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:21.496+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:21.496+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50644/load\": read tcp 127.0.0.1:50651->127.0.0.1:50644: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:21 | 500 | 1.2150414s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:21.523+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:21 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:21 | 200 | 46.2977ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:22.411+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:22.411+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:22.411+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:22.411+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:22.413+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50663" time=2025-10-26T18:52:22.416+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:22.416+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:22.416+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:22.416+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:22.416+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:22.416+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:22.460+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:22.477+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50663" time=2025-10-26T18:52:22.485+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:22.501+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:22.585+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:22.797+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:22.819+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:23.143+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:23.143+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50663/load\": read tcp 127.0.0.1:50670->127.0.0.1:50663: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:23 | 500 | 1.1476022s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:23.170+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:23 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:23 | 200 | 50.5666ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:24.058+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:24.058+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:24.058+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:24.058+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:24.059+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50682" time=2025-10-26T18:52:24.064+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:24.064+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:24.064+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:24.064+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:24.064+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:24.064+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:24.110+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:24.127+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50682" time=2025-10-26T18:52:24.133+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:24.149+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:24.231+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:24.444+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:24.466+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:24.790+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:24.790+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50682/load\": read tcp 127.0.0.1:50689->127.0.0.1:50682: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:24 | 500 | 1.1418636s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:24.818+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:25 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:25 | 200 | 47.6571ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:25.682+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:25.682+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:25.682+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:25.682+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:25.683+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50704" time=2025-10-26T18:52:25.689+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:25.689+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:25.689+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:25.689+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:25.689+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:25.690+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:25.734+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:25.751+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50704" time=2025-10-26T18:52:25.758+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:25.776+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:25.858+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:26.071+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:26.091+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:26.415+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:26.416+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:26.416+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50704/load\": read tcp 127.0.0.1:50711->127.0.0.1:50704: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:26 | 500 | 1.1388094s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:26.443+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:26 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:26 | 200 | 48.6038ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:27.262+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:27.262+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:27.262+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:27.262+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:27.272+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50723" time=2025-10-26T18:52:27.277+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:27.277+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:27.277+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:27.277+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:27.278+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:27.278+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:27.322+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:27.339+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50723" time=2025-10-26T18:52:27.346+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:27.363+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:27.449+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:27.661+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:27.684+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:28.011+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:28.011+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50723/load\": read tcp 127.0.0.1:50730->127.0.0.1:50723: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:28 | 500 | 1.1726113s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:28.045+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:28 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:28 | 200 | 50.6107ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:28.880+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:28.880+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:28.880+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:28.880+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:28.882+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50747" time=2025-10-26T18:52:28.887+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:28.887+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:28.887+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:28.887+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:28.887+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:28.887+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:28.932+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:28.949+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50747" time=2025-10-26T18:52:28.956+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:28.972+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:29.059+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:29.267+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:29.289+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:29.617+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:29.617+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50747/load\": read tcp 127.0.0.1:50754->127.0.0.1:50747: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:29 | 500 | 1.1424097s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:29.646+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 18:52:30 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 18:52:30 | 200 | 57.7387ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T18:52:30.505+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:30.505+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:30.505+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:30.505+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T18:52:30.506+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 50766" time=2025-10-26T18:52:30.510+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T18:52:30.510+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T18:52:30.510+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T18:52:30.510+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T18:52:30.510+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.8 GiB" free_swap="40.2 GiB" time=2025-10-26T18:52:30.510+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T18:52:30.557+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T18:52:30.574+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:50766" time=2025-10-26T18:52:30.579+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:30.595+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v13\ggml-cuda.dll time=2025-10-26T18:52:30.679+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=750,800,860,870,890,900,1000,1100,1200,1210 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T18:52:30.888+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T18:52:30.909+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T18:52:31.232+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T18:52:31.232+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:50766/load\": read tcp 127.0.0.1:50773->127.0.0.1:50766: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 18:52:31 | 500 | 1.1375198s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T18:52:31.260+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" ```
GiteaMirror commented 2026-04-12 21:10:15 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

None of these logs have OLLAMA_LLM_LIBRARY=cuda_v12.

<!-- gh-comment-id:3448734215 --> @rick-github commented on GitHub (Oct 26, 2025): None of these logs have `OLLAMA_LLM_LIBRARY=cuda_v12`.
GiteaMirror commented 2026-04-12 21:10:15 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

weird. let me erase the logs and restart. I will now add it to the windows system environment settings to make shure that it is set.

<!-- gh-comment-id:3448734625 --> @kiliansinger commented on GitHub (Oct 26, 2025): weird. let me erase the logs and restart. I will now add it to the windows system environment settings to make shure that it is set.
GiteaMirror commented 2026-04-12 21:10:16 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

I can now verify that environment is set as proposed in the server.log.
Here a single login session where it failed 10 times and never succeeded:

time=2025-10-26T19:23:50.078+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY:cuda_v12 OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-26T19:23:50.121+01:00 level=INFO source=images.go:522 msg="total blobs: 32"
time=2025-10-26T19:23:50.125+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0"
time=2025-10-26T19:23:50.128+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)"
time=2025-10-26T19:23:50.129+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..."
time=2025-10-26T19:23:50.817+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v12 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.5 GiB"
time=2025-10-26T19:23:50.817+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB"
<!-- gh-comment-id:3448752010 --> @kiliansinger commented on GitHub (Oct 26, 2025): I can now verify that environment is set as proposed in the server.log. Here a single login session where it failed 10 times and never succeeded: ``` time=2025-10-26T19:23:50.078+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY:cuda_v12 OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-26T19:23:50.121+01:00 level=INFO source=images.go:522 msg="total blobs: 32" time=2025-10-26T19:23:50.125+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0" time=2025-10-26T19:23:50.128+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)" time=2025-10-26T19:23:50.129+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..." time=2025-10-26T19:23:50.817+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v12 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.5 GiB" time=2025-10-26T19:23:50.817+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB" ```
GiteaMirror commented 2026-04-12 21:10:16 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

And here with ollama 0.12.6 environment variable set as proposed running through:

time=2025-10-26T19:29:38.738+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY:cuda_v12 OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-26T19:29:38.765+01:00 level=INFO source=images.go:522 msg="total blobs: 32"
time=2025-10-26T19:29:38.767+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0"
time=2025-10-26T19:29:38.768+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)"
time=2025-10-26T19:29:38.769+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..."
time=2025-10-26T19:29:39.876+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v12 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB"
time=2025-10-26T19:29:39.876+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB"
[GIN] 2025/10/26 - 19:29:39 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 19:29:39 | 200 |     35.4967ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T19:29:40.413+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T19:29:40.413+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T19:29:40.413+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T19:29:40.413+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T19:29:40.413+01:00 level=INFO source=server.go:244 msg="using requested gpu library" requested=cuda_v12
time=2025-10-26T19:29:40.415+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 57521"
time=2025-10-26T19:29:40.421+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T19:29:40.422+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T19:29:40.422+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T19:29:40.422+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T19:29:40.422+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.4 GiB" free_swap="45.8 GiB"
time=2025-10-26T19:29:40.422+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T19:29:40.513+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T19:29:40.568+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:57521"
time=2025-10-26T19:29:40.580+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T19:29:40.615+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v12\ggml-cuda.dll
time=2025-10-26T19:29:40.807+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T19:29:41.037+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T19:29:41.086+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T19:29:41.605+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:57521/load\": read tcp 127.0.0.1:57536->127.0.0.1:57521: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/26 - 19:29:41 | 500 |    1.6749044s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-26T19:29:41.635+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
[GIN] 2025/10/26 - 19:29:43 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/26 - 19:29:43 | 200 |     28.8428ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-26T19:29:43.759+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T19:29:43.759+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T19:29:43.759+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T19:29:43.759+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-26T19:29:43.759+01:00 level=INFO source=server.go:244 msg="using requested gpu library" requested=cuda_v12
time=2025-10-26T19:29:43.759+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49780"
time=2025-10-26T19:29:43.764+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-26T19:29:43.764+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-26T19:29:43.764+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-26T19:29:43.764+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-26T19:29:43.764+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="16.9 GiB" free_swap="43.8 GiB"
time=2025-10-26T19:29:43.764+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-26T19:29:43.803+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-26T19:29:43.820+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49780"
time=2025-10-26T19:29:43.830+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T19:29:43.846+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v12\ggml-cuda.dll
time=2025-10-26T19:29:43.937+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-26T19:29:44.059+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T19:29:44.080+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T19:29:44.341+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-26T19:29:44.341+01:00 level=INFO source=ggml.go:480 msg="offloading 28 repeating layers to GPU"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=ggml.go:484 msg="offloading output layer to CPU"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=ggml.go:492 msg="offloaded 28/49 layers to GPU"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=sched.go:482 msg="loaded runners" count=1
time=2025-10-26T19:29:44.342+01:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding"
time=2025-10-26T19:29:44.342+01:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model"
time=2025-10-26T19:29:57.664+01:00 level=INFO source=server.go:1310 msg="llama runner started in 13.90 seconds"
[GIN] 2025/10/26 - 19:29:57 | 200 |   14.1906272s |       127.0.0.1 | POST     "/api/generate"
[GIN] 2025/10/26 - 19:30:06 | 200 |    3.4720297s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2025/10/26 - 19:30:09 | 200 |            0s |       127.0.0.1 | GET      "/api/version"
[GIN] 2025/10/26 - 19:30:09 | 200 |            0s |       127.0.0.1 | GET      "/api/version"
[GIN] 2025/10/26 - 19:30:09 | 404 |      2.6541ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2025/10/26 - 19:30:09 | 200 |     96.7451ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/26 - 19:30:10 | 404 |      2.6244ms |       127.0.0.1 | POST     "/api/show"
<!-- gh-comment-id:3448759729 --> @kiliansinger commented on GitHub (Oct 26, 2025): And here with ollama 0.12.6 environment variable set as proposed running through: ``` time=2025-10-26T19:29:38.738+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY:cuda_v12 OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-26T19:29:38.765+01:00 level=INFO source=images.go:522 msg="total blobs: 32" time=2025-10-26T19:29:38.767+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0" time=2025-10-26T19:29:38.768+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)" time=2025-10-26T19:29:38.769+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..." time=2025-10-26T19:29:39.876+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v12 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.8 GiB" time=2025-10-26T19:29:39.876+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB" [GIN] 2025/10/26 - 19:29:39 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 19:29:39 | 200 | 35.4967ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T19:29:40.413+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T19:29:40.413+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T19:29:40.413+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T19:29:40.413+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T19:29:40.413+01:00 level=INFO source=server.go:244 msg="using requested gpu library" requested=cuda_v12 time=2025-10-26T19:29:40.415+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 57521" time=2025-10-26T19:29:40.421+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T19:29:40.422+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T19:29:40.422+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T19:29:40.422+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T19:29:40.422+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="17.4 GiB" free_swap="45.8 GiB" time=2025-10-26T19:29:40.422+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.8 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T19:29:40.513+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T19:29:40.568+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:57521" time=2025-10-26T19:29:40.580+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T19:29:40.615+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v12\ggml-cuda.dll time=2025-10-26T19:29:40.807+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T19:29:41.037+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T19:29:41.086+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T19:29:41.605+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T19:29:41.605+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:57521/load\": read tcp 127.0.0.1:57536->127.0.0.1:57521: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/26 - 19:29:41 | 500 | 1.6749044s | 127.0.0.1 | POST "/api/generate" time=2025-10-26T19:29:41.635+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" [GIN] 2025/10/26 - 19:29:43 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/26 - 19:29:43 | 200 | 28.8428ms | 127.0.0.1 | POST "/api/show" time=2025-10-26T19:29:43.759+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T19:29:43.759+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T19:29:43.759+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T19:29:43.759+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-26T19:29:43.759+01:00 level=INFO source=server.go:244 msg="using requested gpu library" requested=cuda_v12 time=2025-10-26T19:29:43.759+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 49780" time=2025-10-26T19:29:43.764+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-26T19:29:43.764+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-26T19:29:43.764+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-26T19:29:43.764+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-26T19:29:43.764+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="16.9 GiB" free_swap="43.8 GiB" time=2025-10-26T19:29:43.764+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-26T19:29:43.803+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-26T19:29:43.820+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:49780" time=2025-10-26T19:29:43.830+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T19:29:43.846+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v12\ggml-cuda.dll time=2025-10-26T19:29:43.937+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-26T19:29:44.059+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T19:29:44.080+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T19:29:44.341+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-26T19:29:44.341+01:00 level=INFO source=ggml.go:480 msg="offloading 28 repeating layers to GPU" time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-26T19:29:44.342+01:00 level=INFO source=ggml.go:484 msg="offloading output layer to CPU" time=2025-10-26T19:29:44.342+01:00 level=INFO source=ggml.go:492 msg="offloaded 28/49 layers to GPU" time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-26T19:29:44.342+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-26T19:29:44.342+01:00 level=INFO source=sched.go:482 msg="loaded runners" count=1 time=2025-10-26T19:29:44.342+01:00 level=INFO source=server.go:1272 msg="waiting for llama runner to start responding" time=2025-10-26T19:29:44.342+01:00 level=INFO source=server.go:1306 msg="waiting for server to become available" status="llm server loading model" time=2025-10-26T19:29:57.664+01:00 level=INFO source=server.go:1310 msg="llama runner started in 13.90 seconds" [GIN] 2025/10/26 - 19:29:57 | 200 | 14.1906272s | 127.0.0.1 | POST "/api/generate" [GIN] 2025/10/26 - 19:30:06 | 200 | 3.4720297s | 127.0.0.1 | POST "/api/chat" [GIN] 2025/10/26 - 19:30:09 | 200 | 0s | 127.0.0.1 | GET "/api/version" [GIN] 2025/10/26 - 19:30:09 | 200 | 0s | 127.0.0.1 | GET "/api/version" [GIN] 2025/10/26 - 19:30:09 | 404 | 2.6541ms | 127.0.0.1 | POST "/api/show" [GIN] 2025/10/26 - 19:30:09 | 200 | 96.7451ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/26 - 19:30:10 | 404 | 2.6244ms | 127.0.0.1 | POST "/api/show" ```
GiteaMirror commented 2026-04-12 21:10:16 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

This might be a coincidence but my system also starts the ollama gui for chatting. And it seems that the loading works in more cases shortly before the loading being finished.

<!-- gh-comment-id:3448761380 --> @kiliansinger commented on GitHub (Oct 26, 2025): This might be a coincidence but my system also starts the ollama gui for chatting. And it seems that the loading works in more cases shortly before the loading being finished.
GiteaMirror commented 2026-04-12 21:10:17 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

No it is just a coincidence.

<!-- gh-comment-id:3448762468 --> @kiliansinger commented on GitHub (Oct 26, 2025): No it is just a coincidence.
GiteaMirror commented 2026-04-12 21:10:17 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

but it seems to be more reliable when I wait until the ollama icon appeared in the system tray. But if I do ollama stop ... ollama run... then after it was running once it would never work again.

<!-- gh-comment-id:3448767753 --> @kiliansinger commented on GitHub (Oct 26, 2025): but it seems to be more reliable when I wait until the ollama icon appeared in the system tray. But if I do ``ollama stop ... ollama run...`` then after it was running once it would never work again.
GiteaMirror commented 2026-04-12 21:10:18 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

And if I wait too long then it also does not work. Ideally just when the icon loaded. And of course ollama ps is empty for all those tests. So no other app loads models.

<!-- gh-comment-id:3448768831 --> @kiliansinger commented on GitHub (Oct 26, 2025): And if I wait too long then it also does not work. Ideally just when the icon loaded. And of course ``ollama ps`` is empty for all those tests. So no other app loads models.
GiteaMirror commented 2026-04-12 21:10:18 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

Okay it works in all cases when I do ollama run BIGMODEL right after the ollama icon is loaded where the up-arrow icon in the notification area allows one to see the symbols...

<!-- gh-comment-id:3448786743 --> @kiliansinger commented on GitHub (Oct 26, 2025): Okay it works in all cases when I do ``ollama run BIGMODEL`` right after the ollama icon is loaded where the up-arrow icon in the notification area allows one to see the symbols...
GiteaMirror commented 2026-04-12 21:10:19 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

I also tried to set OLLAMA_LLM_LIBRARY=cuda_v13 but it does not make a difference. Only a few seconds after the ollama icon is loaded to the notification area you always get a fail of loading big models. And small models still work always.

<!-- gh-comment-id:3448828312 --> @kiliansinger commented on GitHub (Oct 26, 2025): I also tried to set OLLAMA_LLM_LIBRARY=cuda_v13 but it does not make a difference. Only a few seconds after the ollama icon is loaded to the notification area you always get a fail of loading big models. And small models still work always.
GiteaMirror commented 2026-04-12 21:10:19 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

If I do not use the environment variable being set then I can still get it to run if I start ollama run BIGMODEL within a few second interval after the ollama icon appears.

<!-- gh-comment-id:3448832519 --> @kiliansinger commented on GitHub (Oct 26, 2025): If I do not use the environment variable being set then I can still get it to run if I start ``ollama run BIGMODEL`` within a few second interval after the ollama icon appears.
GiteaMirror commented 2026-04-12 21:10:19 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

Maybe also of relevance I have also AMD GPU in the laptop due to the CPU:
AMD Ryzen AI 9 HX 370 w/ Radeon 890M

<!-- gh-comment-id:3448851498 --> @kiliansinger commented on GitHub (Oct 26, 2025): Maybe also of relevance I have also AMD GPU in the laptop due to the CPU: AMD Ryzen AI 9 HX 370 w/ Radeon 890M
GiteaMirror commented 2026-04-12 21:10:21 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 26, 2025):

But I think it does not matter because problem persists if I remove the rocm folder. Also if I remove the cuda_12 and cuda_13 folder it is forced to use cpu and loads.

<!-- gh-comment-id:3448857933 --> @kiliansinger commented on GitHub (Oct 26, 2025): But I think it does not matter because problem persists if I remove the rocm folder. Also if I remove the cuda_12 and cuda_13 folder it is forced to use cpu and loads.
GiteaMirror commented 2026-04-12 21:10:21 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

Try rolling back the drivers to v12: https://developer.nvidia.com/cuda-12-9-0-download-archive

<!-- gh-comment-id:3449019375 --> @rick-github commented on GitHub (Oct 26, 2025): Try rolling back the drivers to v12: https://developer.nvidia.com/cuda-12-9-0-download-archive
GiteaMirror commented 2026-04-12 21:10:21 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 27, 2025):

did not improve.
log:

time=2025-10-27T08:16:05.797+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY:cuda_v12 OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]"
time=2025-10-27T08:16:05.830+01:00 level=INFO source=images.go:522 msg="total blobs: 29"
time=2025-10-27T08:16:05.832+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0"
time=2025-10-27T08:16:05.833+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)"
time=2025-10-27T08:16:05.834+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..."
time=2025-10-27T08:16:06.401+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v12 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.5 GiB"
time=2025-10-27T08:16:06.401+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB"
[GIN] 2025/10/27 - 08:16:33 | 200 |       528.1µs |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/27 - 08:16:33 | 200 |      3.3056ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2025/10/27 - 08:16:40 | 200 |            0s |       127.0.0.1 | HEAD     "/"
[GIN] 2025/10/27 - 08:16:40 | 200 |     50.4931ms |       127.0.0.1 | POST     "/api/show"
time=2025-10-27T08:16:40.878+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-27T08:16:40.878+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-27T08:16:40.878+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-27T08:16:40.878+01:00 level=INFO source=server.go:216 msg="enabling flash attention"
time=2025-10-27T08:16:40.878+01:00 level=INFO source=server.go:244 msg="using requested gpu library" requested=cuda_v12
time=2025-10-27T08:16:40.879+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59617"
time=2025-10-27T08:16:40.884+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1
time=2025-10-27T08:16:40.884+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1
time=2025-10-27T08:16:40.884+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1
time=2025-10-27T08:16:40.884+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24
time=2025-10-27T08:16:40.884+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.4 GiB" free_swap="39.9 GiB"
time=2025-10-27T08:16:40.885+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B"
time=2025-10-27T08:16:40.932+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine"
time=2025-10-27T08:16:40.949+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59617"
time=2025-10-27T08:16:40.954+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-27T08:16:40.970+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48
load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248
load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v12\ggml-cuda.dll
time=2025-10-27T08:16:41.072+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang)
time=2025-10-27T08:16:41.217+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
time=2025-10-27T08:16:41.241+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB"
time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB"
time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB"
time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB"
time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB"
time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB"
time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB"
time=2025-10-27T08:16:41.688+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59617/load\": read tcp 127.0.0.1:59624->127.0.0.1:59617: wsarecv: An existing connection was forcibly closed by the remote host."
[GIN] 2025/10/27 - 08:16:41 | 500 |    1.2981685s |       127.0.0.1 | POST     "/api/generate"
time=2025-10-27T08:16:41.724+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409"
<!-- gh-comment-id:3449832821 --> @kiliansinger commented on GitHub (Oct 27, 2025): did not improve. log: ``` time=2025-10-27T08:16:05.797+01:00 level=INFO source=routes.go:1511 msg="server config" env="map[CUDA_VISIBLE_DEVICES:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:INFO OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY:cuda_v12 OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:C:\\Users\\k\\.ollama\\models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[moz-extension://* http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES:]" time=2025-10-27T08:16:05.830+01:00 level=INFO source=images.go:522 msg="total blobs: 29" time=2025-10-27T08:16:05.832+01:00 level=INFO source=images.go:529 msg="total unused blobs removed: 0" time=2025-10-27T08:16:05.833+01:00 level=INFO source=routes.go:1564 msg="Listening on 127.0.0.1:11434 (version 0.12.6)" time=2025-10-27T08:16:05.834+01:00 level=INFO source=runner.go:80 msg="discovering available GPUs..." time=2025-10-27T08:16:06.401+01:00 level=INFO source=types.go:112 msg="inference compute" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA compute=8.9 name=CUDA0 description="NVIDIA GeForce RTX 4070 Laptop GPU" libdirs=ollama,cuda_v12 driver=13.0 pci_id=c4:00.0 type=discrete total="8.0 GiB" available="7.5 GiB" time=2025-10-27T08:16:06.401+01:00 level=INFO source=routes.go:1605 msg="entering low vram mode" "total vram"="8.0 GiB" threshold="20.0 GiB" [GIN] 2025/10/27 - 08:16:33 | 200 | 528.1µs | 127.0.0.1 | HEAD "/" [GIN] 2025/10/27 - 08:16:33 | 200 | 3.3056ms | 127.0.0.1 | GET "/api/tags" [GIN] 2025/10/27 - 08:16:40 | 200 | 0s | 127.0.0.1 | HEAD "/" [GIN] 2025/10/27 - 08:16:40 | 200 | 50.4931ms | 127.0.0.1 | POST "/api/show" time=2025-10-27T08:16:40.878+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-27T08:16:40.878+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-27T08:16:40.878+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-27T08:16:40.878+01:00 level=INFO source=server.go:216 msg="enabling flash attention" time=2025-10-27T08:16:40.878+01:00 level=INFO source=server.go:244 msg="using requested gpu library" requested=cuda_v12 time=2025-10-27T08:16:40.879+01:00 level=INFO source=server.go:400 msg="starting runner" cmd="C:\\Users\\k\\AppData\\Local\\Programs\\Ollama\\ollama.exe runner --ollama-engine --model C:\\Users\\k\\.ollama\\models\\blobs\\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c --port 59617" time=2025-10-27T08:16:40.884+01:00 level=INFO source=server.go:676 msg="loading model" "model layers"=49 requested=-1 time=2025-10-27T08:16:40.884+01:00 level=INFO source=cpu_windows.go:139 msg=packages count=1 time=2025-10-27T08:16:40.884+01:00 level=INFO source=cpu_windows.go:155 msg="efficiency cores detected" maxEfficiencyClass=1 time=2025-10-27T08:16:40.884+01:00 level=INFO source=cpu_windows.go:186 msg="" package=0 cores=12 efficiency=8 threads=24 time=2025-10-27T08:16:40.884+01:00 level=INFO source=server.go:682 msg="system memory" total="31.1 GiB" free="14.4 GiB" free_swap="39.9 GiB" time=2025-10-27T08:16:40.885+01:00 level=INFO source=server.go:690 msg="gpu memory" id=GPU-0f18890a-f457-6c11-0db5-d548aad2c248 library=CUDA available="7.3 GiB" free="7.7 GiB" minimum="457.0 MiB" overhead="0 B" time=2025-10-27T08:16:40.932+01:00 level=INFO source=runner.go:1332 msg="starting ollama engine" time=2025-10-27T08:16:40.949+01:00 level=INFO source=runner.go:1367 msg="Server listening on 127.0.0.1:59617" time=2025-10-27T08:16:40.954+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:49[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:49(0..48)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-27T08:16:40.970+01:00 level=INFO source=ggml.go:134 msg="" architecture=qwen3moe file_type=unknown name=Qwen3-Coder-30B-A3B-Instruct-1M description="" num_tensors=579 num_key_values=48 load_backend: loaded CPU backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\ggml-cpu-icelake.dll ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4070 Laptop GPU, compute capability 8.9, VMM: yes, ID: GPU-0f18890a-f457-6c11-0db5-d548aad2c248 load_backend: loaded CUDA backend from C:\Users\k\AppData\Local\Programs\Ollama\lib\ollama\cuda_v12\ggml-cuda.dll time=2025-10-27T08:16:41.072+01:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX2=1 CPU.0.F16C=1 CPU.0.FMA=1 CPU.0.BMI2=1 CPU.0.AVX512=1 CPU.0.AVX512_VBMI=1 CPU.0.AVX512_VNNI=1 CPU.0.LLAMAFILE=1 CPU.1.LLAMAFILE=1 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,870,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(clang) time=2025-10-27T08:16:41.217+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" time=2025-10-27T08:16:41.241+01:00 level=INFO source=runner.go:1205 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:28[ID:GPU-0f18890a-f457-6c11-0db5-d548aad2c248 Layers:28(20..47)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:206 msg="model weights" device=CUDA0 size="6.9 GiB" time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:211 msg="model weights" device=CPU size="5.0 GiB" time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:217 msg="kv cache" device=CUDA0 size="224.0 MiB" time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:222 msg="kv cache" device=CPU size="160.0 MiB" time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:228 msg="compute graph" device=CUDA0 size="116.5 MiB" time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:233 msg="compute graph" device=CPU size="84.0 MiB" time=2025-10-27T08:16:41.688+01:00 level=INFO source=device.go:238 msg="total memory" size="12.6 GiB" time=2025-10-27T08:16:41.688+01:00 level=INFO source=sched.go:450 msg="Load failed" model=C:\Users\k\.ollama\models\blobs\sha256-475e986fa86970b37cc313a6c32cbeeb64ca1045ae8e3c9c60304437dcb0fe8c error="do load request: Post \"http://127.0.0.1:59617/load\": read tcp 127.0.0.1:59624->127.0.0.1:59617: wsarecv: An existing connection was forcibly closed by the remote host." [GIN] 2025/10/27 - 08:16:41 | 500 | 1.2981685s | 127.0.0.1 | POST "/api/generate" time=2025-10-27T08:16:41.724+01:00 level=ERROR source=server.go:426 msg="llama runner terminated" error="exit status 0xc0000409" ```
GiteaMirror commented 2026-04-12 21:10:23 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 27, 2025):

the cuda error is weird as the folder c:\a does not exist:

CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)

ollama is installed here: C:\Users\k\AppData\Local\Programs\Ollama\ollama.exe

<!-- gh-comment-id:3449839983 --> @kiliansinger commented on GitHub (Oct 27, 2025): the cuda error is weird as the folder c:\a does not exist: ``` CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) ``` ollama is installed here: C:\Users\k\AppData\Local\Programs\Ollama\ollama.exe
GiteaMirror commented 2026-04-12 21:10:23 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 27, 2025):

I want to try to do git bisect. So following docs/development.md I got all prerequisite working and did successful compile. But I cannot find a ollama.exe file.
It generates a lot of dll files. But how can use them. also go run . serve does not do much.

<!-- gh-comment-id:3453113901 --> @kiliansinger commented on GitHub (Oct 27, 2025): I want to try to do ``git bisect``. So following docs/development.md I got all prerequisite working and did successful compile. But I cannot find a ollama.exe file. It generates a lot of dll files. But how can use them. also ``go run . serve`` does not do much.
GiteaMirror commented 2026-04-12 21:10:24 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 27, 2025):

I guess best I look at the workflow...seems to be using a script at scripts/build_windows.ps1 and seems to use msys

<!-- gh-comment-id:3453242267 --> @kiliansinger commented on GitHub (Oct 27, 2025): I guess best I look at the workflow...seems to be using a script at scripts/build_windows.ps1 and seems to use msys
GiteaMirror commented 2026-04-12 21:10:25 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 27, 2025):

I am able to compile and generate fully ollama 0.12.6.
I used newest cuda 13 and compiled also only cuda 13.
need to set in system environment variables: OLLAMA_LLM_LIBRARY = cuda_v13
I first thought it runs better but it actually has the same rates of failures of about 90%. It just happened that right after compiling and installing it worked three times in a row.

<!-- gh-comment-id:3453469808 --> @kiliansinger commented on GitHub (Oct 27, 2025): I am able to compile and generate fully ollama 0.12.6. I used newest cuda 13 and compiled also only cuda 13. need to set in system environment variables: OLLAMA_LLM_LIBRARY = cuda_v13 I first thought it runs better but it actually has the same rates of failures of about 90%. It just happened that right after compiling and installing it worked three times in a row.
GiteaMirror commented 2026-04-12 21:10:26 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 29, 2025):

finally done with git bisect here is the first commit that introduced the regression bug:

3d0b1734c006798960a56acb0ea23ea57e0dd1d9 is the first bad commit
commit 3d0b1734c006798960a56acb0ea23ea57e0dd1d9
Author: Jesse Gross <jesse@ollama.com>
Date:   Tue Sep 9 16:17:31 2025 -0700

    ggml: Preallocate CUDA pool memory

    The GGML CUDA backend allocates additional memory for intermediate
    results during calculation. This memory isn't currently allocated
    during worst case graph reservation and therefore not included in
    scheduling. This means that as these buffers potentially grow
    with context length, we could crash.

    This extends the memory allocation system down layer from the GGML
    graph to the CUDA layer, preallocating the worst case memory there
    as well.

    Fixes #11753

 llama/patches/0022-ggml-No-alloc-mode.patch     | 619 ++++++++++++++++++++++--
 ml/backend/ggml/ggml.go                         |  88 ++--
 ml/backend/ggml/ggml/include/ggml-backend.h     |   2 +-
 ml/backend/ggml/ggml/src/ggml-backend-impl.h    |  14 +
 ml/backend/ggml/ggml/src/ggml-backend.cpp       |  63 ++-
 ml/backend/ggml/ggml/src/ggml-cuda/common.cuh   |  48 +-
 ml/backend/ggml/ggml/src/ggml-cuda/ggml-cuda.cu | 217 +++++++--
 7 files changed, 926 insertions(+), 125 deletions(-)
<!-- gh-comment-id:3462819375 --> @kiliansinger commented on GitHub (Oct 29, 2025): finally done with git bisect here is the first commit that introduced the regression bug: ``` 3d0b1734c006798960a56acb0ea23ea57e0dd1d9 is the first bad commit commit 3d0b1734c006798960a56acb0ea23ea57e0dd1d9 Author: Jesse Gross <jesse@ollama.com> Date: Tue Sep 9 16:17:31 2025 -0700 ggml: Preallocate CUDA pool memory The GGML CUDA backend allocates additional memory for intermediate results during calculation. This memory isn't currently allocated during worst case graph reservation and therefore not included in scheduling. This means that as these buffers potentially grow with context length, we could crash. This extends the memory allocation system down layer from the GGML graph to the CUDA layer, preallocating the worst case memory there as well. Fixes #11753 llama/patches/0022-ggml-No-alloc-mode.patch | 619 ++++++++++++++++++++++-- ml/backend/ggml/ggml.go | 88 ++-- ml/backend/ggml/ggml/include/ggml-backend.h | 2 +- ml/backend/ggml/ggml/src/ggml-backend-impl.h | 14 + ml/backend/ggml/ggml/src/ggml-backend.cpp | 63 ++- ml/backend/ggml/ggml/src/ggml-cuda/common.cuh | 48 +- ml/backend/ggml/ggml/src/ggml-cuda/ggml-cuda.cu | 217 +++++++-- 7 files changed, 926 insertions(+), 125 deletions(-) ```
GiteaMirror commented 2026-04-12 21:10:26 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 29, 2025):

a side note the above code changes correlate well with the log error message:

CUDA error: invalid argument
  current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536
  cuMemUnmap(pool_addr, pool_size)
C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error
<!-- gh-comment-id:3462920157 --> @kiliansinger commented on GitHub (Oct 29, 2025): a side note the above code changes correlate well with the log error message: ``` CUDA error: invalid argument current device: 0, in function ~ggml_cuda_pool_vmm at C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:536 cuMemUnmap(pool_addr, pool_size) C:\a\ollama\ollama\ml\backend\ggml\ggml\src\ggml-cuda\ggml-cuda.cu:88: CUDA error ```
GiteaMirror commented 2026-04-12 21:10:28 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 30, 2025):

The probabilistic behaviour I noticed and described above could mean it is a race condition. My suspicion is raced by:

bool reserving_graph = false;

which is written and read by multiple functions that could run in parallel.

<!-- gh-comment-id:3466395892 --> @kiliansinger commented on GitHub (Oct 30, 2025): The probabilistic behaviour I noticed and described above could mean it is a race condition. My suspicion is raced by: bool reserving_graph = false; which is written and read by multiple functions that could run in parallel.
GiteaMirror commented 2026-04-12 21:10:29 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 30, 2025):

I found uninitalized variable in gglm-cuda.cu line 656:
static std::atomic ggml_cuda_lock_counter;
but setting it to zero also did not fix the issue.

<!-- gh-comment-id:3469752447 --> @kiliansinger commented on GitHub (Oct 30, 2025): I found uninitalized variable in gglm-cuda.cu line 656: static std::atomic<int> ggml_cuda_lock_counter; but setting it to zero also did not fix the issue.
GiteaMirror commented 2026-04-12 21:10:30 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 30, 2025):

I can get it 100% running if I force the alloc value to true.
So the runner\ollamarunner\runner.go line
AllocMemory: true// req.Operation != llm.LoadOperationFit

So my conclusion is that the alloc=false fails. But weirdly sometimes it works.

<!-- gh-comment-id:3470437427 --> @kiliansinger commented on GitHub (Oct 30, 2025): I can get it 100% running if I force the alloc value to true. So the runner\ollamarunner\runner.go line AllocMemory: true// req.Operation != llm.LoadOperationFit So my conclusion is that the alloc=false fails. But weirdly sometimes it works.
GiteaMirror commented 2026-04-12 21:10:30 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Oct 30, 2025):

yes this worked: see PR https://github.com/ollama/ollama/pull/12856

<!-- gh-comment-id:3470508819 --> @kiliansinger commented on GitHub (Oct 30, 2025): yes this worked: see PR https://github.com/ollama/ollama/pull/12856
GiteaMirror commented 2026-04-12 21:10:31 -05:00
Author
Owner
Copy Link

@kiliansinger commented on GitHub (Nov 2, 2025):

Version 0.12.9 solves this issue.

<!-- gh-comment-id:3476988082 --> @kiliansinger commented on GitHub (Nov 2, 2025): Version 0.12.9 solves this issue.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug needs more info
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#8478
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?