github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #11723] ollama 0.11 not using GPU despite detecting them and finding the libraries #69820

New Issue
Closed
opened 2026-05-04 19:28:44 -05:00 by GiteaMirror · 11 comments
GiteaMirror commented 2026-05-04 19:28:44 -05:00
Owner
Copy Link

Originally created by @timbmg on GitHub (Aug 6, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/11723

What is the issue?

I just upgraded from 0.9.6 to 0.11.3 (also tried the 0.11.2). However, it seems none of the models is able to use the GPU (tried with gpt-oss:20b and gemma3:12b-it-fp16). Oddly, the logs show that it does detect the GPUs and finds the libraries, but still it does not use them:

time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:367 msg="offloading 0 repeating layers to GPU"
time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:371 msg="offloading output layer to CPU"
time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:378 msg="offloaded 0/25 layers to GPU"

I also tried with just using 1 GPU; same issue remains.

Relevant log output

Wed Aug  6 09:31:52 2025       
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.129.03             Driver Version: 535.129.03   CUDA Version: 12.2     |
|-----------------------------------------+----------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
|                                         |                      |               MIG M. |
|=========================================+======================+======================|
|   0  NVIDIA A100-SXM4-80GB          On  | 00000000:87:00.0 Off |                    0 |
| N/A   36C    P0              67W / 400W |      4MiB / 81920MiB |      0%      Default |
|                                         |                      |             Disabled |
+-----------------------------------------+----------------------+----------------------+
|   1  NVIDIA A100-SXM4-80GB          On  | 00000000:90:00.0 Off |                    0 |
| N/A   47C    P0              72W / 400W |      4MiB / 81920MiB |      0%      Default |
|                                         |                      |             Disabled |
+-----------------------------------------+----------------------+----------------------+
                                                                                         
+---------------------------------------------------------------------------------------+
| Processes:                                                                            |
|  GPU   GI   CI        PID   Type   Process name                            GPU Memory |
|        ID   ID                                                             Usage      |
|=======================================================================================|
|  No running processes found                                                           |
+---------------------------------------------------------------------------------------+
time=2025-08-06T09:31:52.490+02:00 level=INFO source=routes.go:1297 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0,1 GPU_DEVICE_ORDINAL:0,1 HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:DEBUG OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:10m0s OLLAMA_MAX_LOADED_MODELS:1 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/lab-storage-1/timbmg/.ollama/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[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:0,1 http_proxy: https_proxy: no_proxy:]"
time=2025-08-06T09:31:52.568+02:00 level=INFO source=images.go:477 msg="total blobs: 51"
time=2025-08-06T09:31:52.610+02:00 level=INFO source=images.go:484 msg="total unused blobs removed: 0"
time=2025-08-06T09:31:52.629+02:00 level=INFO source=routes.go:1350 msg="Listening on [::]:11434 (version 0.11.3)"
time=2025-08-06T09:31:52.632+02:00 level=DEBUG source=sched.go:106 msg="starting llm scheduler"
time=2025-08-06T09:31:52.636+02:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs"
time=2025-08-06T09:31:52.657+02:00 level=DEBUG source=gpu.go:98 msg="searching for GPU discovery libraries for NVIDIA"
time=2025-08-06T09:31:52.660+02:00 level=DEBUG source=gpu.go:501 msg="Searching for GPU library" name=libcuda.so*
time=2025-08-06T09:31:52.661+02:00 level=DEBUG source=gpu.go:525 msg="gpu library search" globs="[/mnt/beegfs/work/timbmg/ollama/lib/ollama/libcuda.so* /lab-storage-1/timbmg/ollama-server-slurm/libcuda.so* /usr/local/cuda*/targets/*/lib/libcuda.so* /usr/lib/*-linux-gnu/nvidia/current/libcuda.so* /usr/lib/*-linux-gnu/libcuda.so* /usr/lib/wsl/lib/libcuda.so* /usr/lib/wsl/drivers/*/libcuda.so* /opt/cuda/lib*/libcuda.so* /usr/local/cuda/lib*/libcuda.so* /usr/lib*/libcuda.so* /usr/local/lib*/libcuda.so*]"
time=2025-08-06T09:31:52.666+02:00 level=DEBUG source=gpu.go:558 msg="discovered GPU libraries" paths=[/usr/lib64/libcuda.so.535.129.03]
initializing /usr/lib64/libcuda.so.535.129.03
dlsym: cuInit - 0x7f974a656660
dlsym: cuDriverGetVersion - 0x7f974a656680
dlsym: cuDeviceGetCount - 0x7f974a6566c0
dlsym: cuDeviceGet - 0x7f974a6566a0
dlsym: cuDeviceGetAttribute - 0x7f974a6567a0
dlsym: cuDeviceGetUuid - 0x7f974a656700
dlsym: cuDeviceGetName - 0x7f974a6566e0
dlsym: cuCtxCreate_v3 - 0x7f974a65e360
dlsym: cuMemGetInfo_v2 - 0x7f974a669850
dlsym: cuCtxDestroy - 0x7f974a6b8940
calling cuInit
calling cuDriverGetVersion
raw version 0x2ef4
CUDA driver version: 12.2
calling cuDeviceGetCount
device count 2
time=2025-08-06T09:31:53.549+02:00 level=DEBUG source=gpu.go:125 msg="detected GPUs" count=2 library=/usr/lib64/libcuda.so.535.129.03
[GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f] CUDA totalMem 81050mb
[GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f] CUDA freeMem 80623mb
[GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f] Compute Capability 8.0
[GPU-c4881392-6d2b-de2e-82ef-835f242bd71c] CUDA totalMem 81050mb
[GPU-c4881392-6d2b-de2e-82ef-835f242bd71c] CUDA freeMem 80623mb
[GPU-c4881392-6d2b-de2e-82ef-835f242bd71c] Compute Capability 8.0
time=2025-08-06T09:31:54.406+02:00 level=DEBUG source=amd_linux.go:419 msg="amdgpu driver not detected /sys/module/amdgpu"
releasing cuda driver library
time=2025-08-06T09:31:54.410+02:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f library=cuda variant=v12 compute=8.0 driver=12.2 name="NVIDIA A100-SXM4-80GB" total="79.2 GiB" available="78.7 GiB"
time=2025-08-06T09:31:54.411+02:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-c4881392-6d2b-de2e-82ef-835f242bd71c library=cuda variant=v12 compute=8.0 driver=12.2 name="NVIDIA A100-SXM4-80GB" total="79.2 GiB" available="78.7 GiB"
[GIN] 2025/08/06 - 09:33:10 | 200 |    2.360317ms |    10.167.11.14 | HEAD     "/"
time=2025-08-06T09:33:10.953+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32
[GIN] 2025/08/06 - 09:33:10 | 200 |  267.061651ms |    10.167.11.14 | POST     "/api/show"
time=2025-08-06T09:33:11.543+02:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="2015.7 GiB" before.free="1928.5 GiB" before.free_swap="0 B" now.total="2015.7 GiB" now.free="1928.9 GiB" now.free_swap="0 B"
initializing /usr/lib64/libcuda.so.535.129.03
dlsym: cuInit - 0x7f974a656660
dlsym: cuDriverGetVersion - 0x7f974a656680
dlsym: cuDeviceGetCount - 0x7f974a6566c0
dlsym: cuDeviceGet - 0x7f974a6566a0
dlsym: cuDeviceGetAttribute - 0x7f974a6567a0
dlsym: cuDeviceGetUuid - 0x7f974a656700
dlsym: cuDeviceGetName - 0x7f974a6566e0
dlsym: cuCtxCreate_v3 - 0x7f974a65e360
dlsym: cuMemGetInfo_v2 - 0x7f974a669850
dlsym: cuCtxDestroy - 0x7f974a6b8940
calling cuInit
calling cuDriverGetVersion
raw version 0x2ef4
CUDA driver version: 12.2
calling cuDeviceGetCount
device count 2
time=2025-08-06T09:33:12.014+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB"
time=2025-08-06T09:33:12.422+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-c4881392-6d2b-de2e-82ef-835f242bd71c name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB"
releasing cuda driver library
time=2025-08-06T09:33:12.492+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32
time=2025-08-06T09:33:12.616+02:00 level=DEBUG source=sched.go:226 msg="loading first model" model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583
time=2025-08-06T09:33:12.618+02:00 level=DEBUG source=memory.go:111 msg=evaluating library=cuda gpu_count=1 available="[78.7 GiB]"
time=2025-08-06T09:33:12.620+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=gptoss.vision.block_count default=0
time=2025-08-06T09:33:12.623+02:00 level=INFO source=sched.go:786 msg="new model will fit in available VRAM in single GPU, loading" model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 gpu=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f parallel=1 available=84539604992 required="14.9 GiB"
time=2025-08-06T09:33:12.624+02:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="2015.7 GiB" before.free="1928.9 GiB" before.free_swap="0 B" now.total="2015.7 GiB" now.free="1928.3 GiB" now.free_swap="0 B"
initializing /usr/lib64/libcuda.so.535.129.03
dlsym: cuInit - 0x7f974a656660
dlsym: cuDriverGetVersion - 0x7f974a656680
dlsym: cuDeviceGetCount - 0x7f974a6566c0
dlsym: cuDeviceGet - 0x7f974a6566a0
dlsym: cuDeviceGetAttribute - 0x7f974a6567a0
dlsym: cuDeviceGetUuid - 0x7f974a656700
dlsym: cuDeviceGetName - 0x7f974a6566e0
dlsym: cuCtxCreate_v3 - 0x7f974a65e360
dlsym: cuMemGetInfo_v2 - 0x7f974a669850
dlsym: cuCtxDestroy - 0x7f974a6b8940
calling cuInit
calling cuDriverGetVersion
raw version 0x2ef4
CUDA driver version: 12.2
calling cuDeviceGetCount
device count 2
time=2025-08-06T09:33:13.049+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB"
time=2025-08-06T09:33:13.526+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-c4881392-6d2b-de2e-82ef-835f242bd71c name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB"
releasing cuda driver library
time=2025-08-06T09:33:13.529+02:00 level=INFO source=server.go:135 msg="system memory" total="2015.7 GiB" free="1928.3 GiB" free_swap="0 B"
time=2025-08-06T09:33:13.531+02:00 level=DEBUG source=memory.go:111 msg=evaluating library=cuda gpu_count=1 available="[78.7 GiB]"
time=2025-08-06T09:33:13.533+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=gptoss.vision.block_count default=0
time=2025-08-06T09:33:13.534+02:00 level=INFO source=server.go:175 msg=offload library=cuda layers.requested=-1 layers.model=25 layers.offload=25 layers.split="" memory.available="[78.7 GiB]" memory.gpu_overhead="0 B" memory.required.full="14.9 GiB" memory.required.partial="14.9 GiB" memory.required.kv="300.0 MiB" memory.required.allocations="[14.9 GiB]" memory.weights.total="11.7 GiB" memory.weights.repeating="10.7 GiB" memory.weights.nonrepeating="1.1 GiB" memory.graph.full="2.0 GiB" memory.graph.partial="2.0 GiB"
time=2025-08-06T09:33:13.539+02:00 level=DEBUG source=server.go:291 msg="compatible gpu libraries" compatible=[]
time=2025-08-06T09:33:13.645+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32
time=2025-08-06T09:33:13.647+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=tokenizer.ggml.pretokenizer default="[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]*[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]+(?i:'s|'t|'re|'ve|'m|'ll|'d)?|[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]+[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]*(?i:'s|'t|'re|'ve|'m|'ll|'d)?|\\p{N}{1,3}| ?[^\\s\\p{L}\\p{N}]+[\\r\\n/]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+"
time=2025-08-06T09:33:13.652+02:00 level=INFO source=server.go:438 msg="starting llama server" cmd="/mnt/beegfs/work/timbmg/ollama/bin/ollama runner --ollama-engine --model /lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 --ctx-size 8192 --batch-size 512 --n-gpu-layers 25 --threads 128 --parallel 1 --port 35503"
time=2025-08-06T09:33:13.654+02:00 level=DEBUG source=server.go:439 msg=subprocess OLLAMA_MAX_LOADED_MODELS=1 OLLAMA_LOAD_TIMEOUT=10m OLLAMA_DEBUG=1 OLLAMA_HOST=0.0.0.0:11434 ROCR_VISIBLE_DEVICES=0,1 CUDA_VISIBLE_DEVICES=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f PATH=/storage/lab/work/timbmg/.local/share/../bin:/lab-storage-1/timbmg/miniconda/bin:/lab-storage-1/timbmg/miniconda/condabin:/home/timbmg/anaconda3/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/puppetlabs/bin:/lab-storage-1/timbmg/ollama/bin:/lab-storage-1/timbmg/.local/share/kitty-ssh-kitten/kitty/bin GPU_DEVICE_ORDINAL=0,1 OLLAMA_LIBRARY_PATH=/mnt/beegfs/work/timbmg/ollama/lib/ollama LD_LIBRARY_PATH=/mnt/beegfs/work/timbmg/ollama/lib/ollama:/mnt/beegfs/work/timbmg/ollama/lib/ollama
time=2025-08-06T09:33:13.686+02:00 level=INFO source=sched.go:481 msg="loaded runners" count=1
time=2025-08-06T09:33:13.706+02:00 level=INFO source=server.go:598 msg="waiting for llama runner to start responding"
time=2025-08-06T09:33:13.764+02:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server not responding"
time=2025-08-06T09:33:13.813+02:00 level=INFO source=runner.go:925 msg="starting ollama engine"
time=2025-08-06T09:33:13.813+02:00 level=INFO source=runner.go:983 msg="Server listening on 127.0.0.1:35503"
time=2025-08-06T09:33:13.920+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32
time=2025-08-06T09:33:13.922+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.name default=""
time=2025-08-06T09:33:13.922+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.description default=""
time=2025-08-06T09:33:13.923+02:00 level=INFO source=ggml.go:92 msg="" architecture=gptoss file_type=MXFP4 name="" description="" num_tensors=315 num_key_values=30
time=2025-08-06T09:33:13.923+02:00 level=DEBUG source=ggml.go:94 msg="ggml backend load all from path" path=/mnt/beegfs/work/timbmg/ollama/lib/ollama
time=2025-08-06T09:33:14.025+02:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server loading model"
load_backend: loaded CPU backend from /mnt/beegfs/work/timbmg/ollama/lib/ollama/libggml-cpu-haswell.so
time=2025-08-06T09:33:14.063+02:00 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.LLAMAFILE=1 CPU.1.LLAMAFILE=1 compiler=cgo(gcc)
time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:367 msg="offloading 0 repeating layers to GPU"
time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:371 msg="offloading output layer to CPU"
time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:378 msg="offloaded 0/25 layers to GPU"
time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:381 msg="model weights" buffer=CPU size="12.8 GiB"
time=2025-08-06T09:33:14.065+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=tokenizer.ggml.pretokenizer default="[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]*[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]+(?i:'s|'t|'re|'ve|'m|'ll|'d)?|[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]+[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]*(?i:'s|'t|'re|'ve|'m|'ll|'d)?|\\p{N}{1,3}| ?[^\\s\\p{L}\\p{N}]+[\\r\\n/]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+"
time=2025-08-06T09:33:14.178+02:00 level=DEBUG source=ggml.go:654 msg="compute graph" nodes=1847 splits=1
time=2025-08-06T09:33:14.178+02:00 level=INFO source=ggml.go:672 msg="compute graph" backend=CPU buffer_type=CPU size="2.0 GiB"
time=2025-08-06T09:33:14.178+02:00 level=DEBUG source=runner.go:883 msg=memory allocated.InputWeights=1158266880A allocated.CPU.Weights="[477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 1158278400A]" allocated.CPU.Cache="[9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 0U]" allocated.CPU.Graph=2178285568A
time=2025-08-06T09:33:14.277+02:00 level=DEBUG source=server.go:643 msg="model load progress 0.00"
...
time=2025-08-06T09:33:39.541+02:00 level=DEBUG source=server.go:643 msg="model load progress 1.00"
time=2025-08-06T09:33:39.794+02:00 level=INFO source=server.go:637 msg="llama runner started in 26.10 seconds"
time=2025-08-06T09:33:39.795+02:00 level=DEBUG source=sched.go:493 msg="finished setting up" runner.name=registry.ollama.ai/library/gpt-oss:20b runner.inference=cuda runner.devices=1 runner.size="14.9 GiB" runner.vram="14.9 GiB" runner.parallel=1 runner.pid=2068391 runner.model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 runner.num_ctx=8192
[GIN] 2025/08/06 - 09:33:39 | 200 | 28.442849614s |    10.167.11.14 | POST     "/api/generate"
time=2025-08-06T09:33:39.800+02:00 level=DEBUG source=sched.go:501 msg="context for request finished"
time=2025-08-06T09:33:39.801+02:00 level=DEBUG source=sched.go:341 msg="runner with non-zero duration has gone idle, adding timer" runner.name=registry.ollama.ai/library/gpt-oss:20b runner.inference=cuda runner.devices=1 runner.size="14.9 GiB" runner.vram="14.9 GiB" runner.parallel=1 runner.pid=2068391 runner.model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 runner.num_ctx=8192 duration=5m0s
time=2025-08-06T09:33:39.803+02:00 level=DEBUG source=sched.go:359 msg="after processing request finished event" runner.name=registry.ollama.ai/library/gpt-oss:20b runner.inference=cuda runner.devices=1 runner.size="14.9 GiB" runner.vram="14.9 GiB" runner.parallel=1 runner.pid=2068391 runner.model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 runner.num_ctx=8192 refCount=0

OS

Linux

GPU

Nvidia

CPU

No response

Ollama version

0.11.3

Originally created by @timbmg on GitHub (Aug 6, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/11723 ### What is the issue? I just upgraded from 0.9.6 to 0.11.3 (also tried the 0.11.2). However, it seems none of the models is able to use the GPU (tried with gpt-oss:20b and gemma3:12b-it-fp16). Oddly, the logs show that it does detect the GPUs and finds the libraries, but still it does not use them: ``` time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:367 msg="offloading 0 repeating layers to GPU" time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:371 msg="offloading output layer to CPU" time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:378 msg="offloaded 0/25 layers to GPU" ``` I also tried with just using 1 GPU; same issue remains. ### Relevant log output ```shell Wed Aug 6 09:31:52 2025 +---------------------------------------------------------------------------------------+ | NVIDIA-SMI 535.129.03 Driver Version: 535.129.03 CUDA Version: 12.2 | |-----------------------------------------+----------------------+----------------------+ | GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+======================+======================| | 0 NVIDIA A100-SXM4-80GB On | 00000000:87:00.0 Off | 0 | | N/A 36C P0 67W / 400W | 4MiB / 81920MiB | 0% Default | | | | Disabled | +-----------------------------------------+----------------------+----------------------+ | 1 NVIDIA A100-SXM4-80GB On | 00000000:90:00.0 Off | 0 | | N/A 47C P0 72W / 400W | 4MiB / 81920MiB | 0% Default | | | | Disabled | +-----------------------------------------+----------------------+----------------------+ +---------------------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=======================================================================================| | No running processes found | +---------------------------------------------------------------------------------------+ time=2025-08-06T09:31:52.490+02:00 level=INFO source=routes.go:1297 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0,1 GPU_DEVICE_ORDINAL:0,1 HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:DEBUG OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:10m0s OLLAMA_MAX_LOADED_MODELS:1 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/lab-storage-1/timbmg/.ollama/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[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:0,1 http_proxy: https_proxy: no_proxy:]" time=2025-08-06T09:31:52.568+02:00 level=INFO source=images.go:477 msg="total blobs: 51" time=2025-08-06T09:31:52.610+02:00 level=INFO source=images.go:484 msg="total unused blobs removed: 0" time=2025-08-06T09:31:52.629+02:00 level=INFO source=routes.go:1350 msg="Listening on [::]:11434 (version 0.11.3)" time=2025-08-06T09:31:52.632+02:00 level=DEBUG source=sched.go:106 msg="starting llm scheduler" time=2025-08-06T09:31:52.636+02:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs" time=2025-08-06T09:31:52.657+02:00 level=DEBUG source=gpu.go:98 msg="searching for GPU discovery libraries for NVIDIA" time=2025-08-06T09:31:52.660+02:00 level=DEBUG source=gpu.go:501 msg="Searching for GPU library" name=libcuda.so* time=2025-08-06T09:31:52.661+02:00 level=DEBUG source=gpu.go:525 msg="gpu library search" globs="[/mnt/beegfs/work/timbmg/ollama/lib/ollama/libcuda.so* /lab-storage-1/timbmg/ollama-server-slurm/libcuda.so* /usr/local/cuda*/targets/*/lib/libcuda.so* /usr/lib/*-linux-gnu/nvidia/current/libcuda.so* /usr/lib/*-linux-gnu/libcuda.so* /usr/lib/wsl/lib/libcuda.so* /usr/lib/wsl/drivers/*/libcuda.so* /opt/cuda/lib*/libcuda.so* /usr/local/cuda/lib*/libcuda.so* /usr/lib*/libcuda.so* /usr/local/lib*/libcuda.so*]" time=2025-08-06T09:31:52.666+02:00 level=DEBUG source=gpu.go:558 msg="discovered GPU libraries" paths=[/usr/lib64/libcuda.so.535.129.03] initializing /usr/lib64/libcuda.so.535.129.03 dlsym: cuInit - 0x7f974a656660 dlsym: cuDriverGetVersion - 0x7f974a656680 dlsym: cuDeviceGetCount - 0x7f974a6566c0 dlsym: cuDeviceGet - 0x7f974a6566a0 dlsym: cuDeviceGetAttribute - 0x7f974a6567a0 dlsym: cuDeviceGetUuid - 0x7f974a656700 dlsym: cuDeviceGetName - 0x7f974a6566e0 dlsym: cuCtxCreate_v3 - 0x7f974a65e360 dlsym: cuMemGetInfo_v2 - 0x7f974a669850 dlsym: cuCtxDestroy - 0x7f974a6b8940 calling cuInit calling cuDriverGetVersion raw version 0x2ef4 CUDA driver version: 12.2 calling cuDeviceGetCount device count 2 time=2025-08-06T09:31:53.549+02:00 level=DEBUG source=gpu.go:125 msg="detected GPUs" count=2 library=/usr/lib64/libcuda.so.535.129.03 [GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f] CUDA totalMem 81050mb [GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f] CUDA freeMem 80623mb [GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f] Compute Capability 8.0 [GPU-c4881392-6d2b-de2e-82ef-835f242bd71c] CUDA totalMem 81050mb [GPU-c4881392-6d2b-de2e-82ef-835f242bd71c] CUDA freeMem 80623mb [GPU-c4881392-6d2b-de2e-82ef-835f242bd71c] Compute Capability 8.0 time=2025-08-06T09:31:54.406+02:00 level=DEBUG source=amd_linux.go:419 msg="amdgpu driver not detected /sys/module/amdgpu" releasing cuda driver library time=2025-08-06T09:31:54.410+02:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f library=cuda variant=v12 compute=8.0 driver=12.2 name="NVIDIA A100-SXM4-80GB" total="79.2 GiB" available="78.7 GiB" time=2025-08-06T09:31:54.411+02:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-c4881392-6d2b-de2e-82ef-835f242bd71c library=cuda variant=v12 compute=8.0 driver=12.2 name="NVIDIA A100-SXM4-80GB" total="79.2 GiB" available="78.7 GiB" [GIN] 2025/08/06 - 09:33:10 | 200 | 2.360317ms | 10.167.11.14 | HEAD "/" time=2025-08-06T09:33:10.953+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32 [GIN] 2025/08/06 - 09:33:10 | 200 | 267.061651ms | 10.167.11.14 | POST "/api/show" time=2025-08-06T09:33:11.543+02:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="2015.7 GiB" before.free="1928.5 GiB" before.free_swap="0 B" now.total="2015.7 GiB" now.free="1928.9 GiB" now.free_swap="0 B" initializing /usr/lib64/libcuda.so.535.129.03 dlsym: cuInit - 0x7f974a656660 dlsym: cuDriverGetVersion - 0x7f974a656680 dlsym: cuDeviceGetCount - 0x7f974a6566c0 dlsym: cuDeviceGet - 0x7f974a6566a0 dlsym: cuDeviceGetAttribute - 0x7f974a6567a0 dlsym: cuDeviceGetUuid - 0x7f974a656700 dlsym: cuDeviceGetName - 0x7f974a6566e0 dlsym: cuCtxCreate_v3 - 0x7f974a65e360 dlsym: cuMemGetInfo_v2 - 0x7f974a669850 dlsym: cuCtxDestroy - 0x7f974a6b8940 calling cuInit calling cuDriverGetVersion raw version 0x2ef4 CUDA driver version: 12.2 calling cuDeviceGetCount device count 2 time=2025-08-06T09:33:12.014+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB" time=2025-08-06T09:33:12.422+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-c4881392-6d2b-de2e-82ef-835f242bd71c name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB" releasing cuda driver library time=2025-08-06T09:33:12.492+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32 time=2025-08-06T09:33:12.616+02:00 level=DEBUG source=sched.go:226 msg="loading first model" model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 time=2025-08-06T09:33:12.618+02:00 level=DEBUG source=memory.go:111 msg=evaluating library=cuda gpu_count=1 available="[78.7 GiB]" time=2025-08-06T09:33:12.620+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=gptoss.vision.block_count default=0 time=2025-08-06T09:33:12.623+02:00 level=INFO source=sched.go:786 msg="new model will fit in available VRAM in single GPU, loading" model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 gpu=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f parallel=1 available=84539604992 required="14.9 GiB" time=2025-08-06T09:33:12.624+02:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="2015.7 GiB" before.free="1928.9 GiB" before.free_swap="0 B" now.total="2015.7 GiB" now.free="1928.3 GiB" now.free_swap="0 B" initializing /usr/lib64/libcuda.so.535.129.03 dlsym: cuInit - 0x7f974a656660 dlsym: cuDriverGetVersion - 0x7f974a656680 dlsym: cuDeviceGetCount - 0x7f974a6566c0 dlsym: cuDeviceGet - 0x7f974a6566a0 dlsym: cuDeviceGetAttribute - 0x7f974a6567a0 dlsym: cuDeviceGetUuid - 0x7f974a656700 dlsym: cuDeviceGetName - 0x7f974a6566e0 dlsym: cuCtxCreate_v3 - 0x7f974a65e360 dlsym: cuMemGetInfo_v2 - 0x7f974a669850 dlsym: cuCtxDestroy - 0x7f974a6b8940 calling cuInit calling cuDriverGetVersion raw version 0x2ef4 CUDA driver version: 12.2 calling cuDeviceGetCount device count 2 time=2025-08-06T09:33:13.049+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB" time=2025-08-06T09:33:13.526+02:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-c4881392-6d2b-de2e-82ef-835f242bd71c name="NVIDIA A100-SXM4-80GB" overhead="0 B" before.total="79.2 GiB" before.free="78.7 GiB" now.total="79.2 GiB" now.free="78.7 GiB" now.used="427.4 MiB" releasing cuda driver library time=2025-08-06T09:33:13.529+02:00 level=INFO source=server.go:135 msg="system memory" total="2015.7 GiB" free="1928.3 GiB" free_swap="0 B" time=2025-08-06T09:33:13.531+02:00 level=DEBUG source=memory.go:111 msg=evaluating library=cuda gpu_count=1 available="[78.7 GiB]" time=2025-08-06T09:33:13.533+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=gptoss.vision.block_count default=0 time=2025-08-06T09:33:13.534+02:00 level=INFO source=server.go:175 msg=offload library=cuda layers.requested=-1 layers.model=25 layers.offload=25 layers.split="" memory.available="[78.7 GiB]" memory.gpu_overhead="0 B" memory.required.full="14.9 GiB" memory.required.partial="14.9 GiB" memory.required.kv="300.0 MiB" memory.required.allocations="[14.9 GiB]" memory.weights.total="11.7 GiB" memory.weights.repeating="10.7 GiB" memory.weights.nonrepeating="1.1 GiB" memory.graph.full="2.0 GiB" memory.graph.partial="2.0 GiB" time=2025-08-06T09:33:13.539+02:00 level=DEBUG source=server.go:291 msg="compatible gpu libraries" compatible=[] time=2025-08-06T09:33:13.645+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32 time=2025-08-06T09:33:13.647+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=tokenizer.ggml.pretokenizer default="[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]*[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]+(?i:'s|'t|'re|'ve|'m|'ll|'d)?|[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]+[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]*(?i:'s|'t|'re|'ve|'m|'ll|'d)?|\\p{N}{1,3}| ?[^\\s\\p{L}\\p{N}]+[\\r\\n/]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+" time=2025-08-06T09:33:13.652+02:00 level=INFO source=server.go:438 msg="starting llama server" cmd="/mnt/beegfs/work/timbmg/ollama/bin/ollama runner --ollama-engine --model /lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 --ctx-size 8192 --batch-size 512 --n-gpu-layers 25 --threads 128 --parallel 1 --port 35503" time=2025-08-06T09:33:13.654+02:00 level=DEBUG source=server.go:439 msg=subprocess OLLAMA_MAX_LOADED_MODELS=1 OLLAMA_LOAD_TIMEOUT=10m OLLAMA_DEBUG=1 OLLAMA_HOST=0.0.0.0:11434 ROCR_VISIBLE_DEVICES=0,1 CUDA_VISIBLE_DEVICES=GPU-d988422a-a4db-43bb-7c8c-d06cd5df7d1f PATH=/storage/lab/work/timbmg/.local/share/../bin:/lab-storage-1/timbmg/miniconda/bin:/lab-storage-1/timbmg/miniconda/condabin:/home/timbmg/anaconda3/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/puppetlabs/bin:/lab-storage-1/timbmg/ollama/bin:/lab-storage-1/timbmg/.local/share/kitty-ssh-kitten/kitty/bin GPU_DEVICE_ORDINAL=0,1 OLLAMA_LIBRARY_PATH=/mnt/beegfs/work/timbmg/ollama/lib/ollama LD_LIBRARY_PATH=/mnt/beegfs/work/timbmg/ollama/lib/ollama:/mnt/beegfs/work/timbmg/ollama/lib/ollama time=2025-08-06T09:33:13.686+02:00 level=INFO source=sched.go:481 msg="loaded runners" count=1 time=2025-08-06T09:33:13.706+02:00 level=INFO source=server.go:598 msg="waiting for llama runner to start responding" time=2025-08-06T09:33:13.764+02:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server not responding" time=2025-08-06T09:33:13.813+02:00 level=INFO source=runner.go:925 msg="starting ollama engine" time=2025-08-06T09:33:13.813+02:00 level=INFO source=runner.go:983 msg="Server listening on 127.0.0.1:35503" time=2025-08-06T09:33:13.920+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.alignment default=32 time=2025-08-06T09:33:13.922+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.name default="" time=2025-08-06T09:33:13.922+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=general.description default="" time=2025-08-06T09:33:13.923+02:00 level=INFO source=ggml.go:92 msg="" architecture=gptoss file_type=MXFP4 name="" description="" num_tensors=315 num_key_values=30 time=2025-08-06T09:33:13.923+02:00 level=DEBUG source=ggml.go:94 msg="ggml backend load all from path" path=/mnt/beegfs/work/timbmg/ollama/lib/ollama time=2025-08-06T09:33:14.025+02:00 level=INFO source=server.go:632 msg="waiting for server to become available" status="llm server loading model" load_backend: loaded CPU backend from /mnt/beegfs/work/timbmg/ollama/lib/ollama/libggml-cpu-haswell.so time=2025-08-06T09:33:14.063+02:00 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.LLAMAFILE=1 CPU.1.LLAMAFILE=1 compiler=cgo(gcc) time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:367 msg="offloading 0 repeating layers to GPU" time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:371 msg="offloading output layer to CPU" time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:378 msg="offloaded 0/25 layers to GPU" time=2025-08-06T09:33:14.064+02:00 level=INFO source=ggml.go:381 msg="model weights" buffer=CPU size="12.8 GiB" time=2025-08-06T09:33:14.065+02:00 level=DEBUG source=ggml.go:208 msg="key with type not found" key=tokenizer.ggml.pretokenizer default="[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]*[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]+(?i:'s|'t|'re|'ve|'m|'ll|'d)?|[^\\r\\n\\p{L}\\p{N}]?[\\p{Lu}\\p{Lt}\\p{Lm}\\p{Lo}\\p{M}]+[\\p{Ll}\\p{Lm}\\p{Lo}\\p{M}]*(?i:'s|'t|'re|'ve|'m|'ll|'d)?|\\p{N}{1,3}| ?[^\\s\\p{L}\\p{N}]+[\\r\\n/]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+" time=2025-08-06T09:33:14.178+02:00 level=DEBUG source=ggml.go:654 msg="compute graph" nodes=1847 splits=1 time=2025-08-06T09:33:14.178+02:00 level=INFO source=ggml.go:672 msg="compute graph" backend=CPU buffer_type=CPU size="2.0 GiB" time=2025-08-06T09:33:14.178+02:00 level=DEBUG source=runner.go:883 msg=memory allocated.InputWeights=1158266880A allocated.CPU.Weights="[477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 477075584A 1158278400A]" allocated.CPU.Cache="[9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 9437184A 16777216A 0U]" allocated.CPU.Graph=2178285568A time=2025-08-06T09:33:14.277+02:00 level=DEBUG source=server.go:643 msg="model load progress 0.00" ... time=2025-08-06T09:33:39.541+02:00 level=DEBUG source=server.go:643 msg="model load progress 1.00" time=2025-08-06T09:33:39.794+02:00 level=INFO source=server.go:637 msg="llama runner started in 26.10 seconds" time=2025-08-06T09:33:39.795+02:00 level=DEBUG source=sched.go:493 msg="finished setting up" runner.name=registry.ollama.ai/library/gpt-oss:20b runner.inference=cuda runner.devices=1 runner.size="14.9 GiB" runner.vram="14.9 GiB" runner.parallel=1 runner.pid=2068391 runner.model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 runner.num_ctx=8192 [GIN] 2025/08/06 - 09:33:39 | 200 | 28.442849614s | 10.167.11.14 | POST "/api/generate" time=2025-08-06T09:33:39.800+02:00 level=DEBUG source=sched.go:501 msg="context for request finished" time=2025-08-06T09:33:39.801+02:00 level=DEBUG source=sched.go:341 msg="runner with non-zero duration has gone idle, adding timer" runner.name=registry.ollama.ai/library/gpt-oss:20b runner.inference=cuda runner.devices=1 runner.size="14.9 GiB" runner.vram="14.9 GiB" runner.parallel=1 runner.pid=2068391 runner.model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 runner.num_ctx=8192 duration=5m0s time=2025-08-06T09:33:39.803+02:00 level=DEBUG source=sched.go:359 msg="after processing request finished event" runner.name=registry.ollama.ai/library/gpt-oss:20b runner.inference=cuda runner.devices=1 runner.size="14.9 GiB" runner.vram="14.9 GiB" runner.parallel=1 runner.pid=2068391 runner.model=/lab-storage-1/timbmg/.ollama/models/blobs/sha256-b112e727c6f18875636c56a779790a590d705aec9e1c0eb5a97d51fc2a778583 runner.num_ctx=8192 refCount=0 ``` ### OS Linux ### GPU Nvidia ### CPU _No response_ ### Ollama version 0.11.3
GiteaMirror added the bug label 2026-05-04 19:28:44 -05:00
GiteaMirror commented 2026-05-04 19:28:48 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 6, 2025):

load_backend: loaded CPU backend from /mnt/beegfs/work/timbmg/ollama/lib/ollama/libggml-cpu-haswell.so

No GPU backends found. How did you install ollama?

<!-- gh-comment-id:3157976206 --> @rick-github commented on GitHub (Aug 6, 2025): ``` load_backend: loaded CPU backend from /mnt/beegfs/work/timbmg/ollama/lib/ollama/libggml-cpu-haswell.so ``` No GPU backends found. How did you install ollama?
GiteaMirror commented 2026-05-04 19:28:49 -05:00
Author
Owner
Copy Link

@timbmg commented on GitHub (Aug 6, 2025):

I installed it manually like this:

curl -LO https://ollama.com/download/ollama-linux-amd64.tgz
tar -C ./ollama -xzf ollama-linux-amd64.tgz

Actually, get 0.11.3 i used https://github.com/ollama/ollama/releases/download/v0.11.3/ollama-linux-amd64.tgz

<!-- gh-comment-id:3158003856 --> @timbmg commented on GitHub (Aug 6, 2025): I installed it manually like this: ```sh curl -LO https://ollama.com/download/ollama-linux-amd64.tgz tar -C ./ollama -xzf ollama-linux-amd64.tgz ``` Actually, get 0.11.3 i used https://github.com/ollama/ollama/releases/download/v0.11.3/ollama-linux-amd64.tgz
GiteaMirror commented 2026-05-04 19:28:50 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 6, 2025):

The runner is being started as /mnt/beegfs/work/timbmg/ollama/bin/ollama, not /usr/bin/ollama. So it's likely the bin directory was moved/copied but not the entirety of the lib directory.

<!-- gh-comment-id:3158047903 --> @rick-github commented on GitHub (Aug 6, 2025): The runner is being started as /mnt/beegfs/work/timbmg/ollama/bin/ollama, not /usr/bin/ollama. So it's likely the bin directory was moved/copied but not the entirety of the lib directory.
GiteaMirror commented 2026-05-04 19:28:51 -05:00
Author
Owner
Copy Link

@timbmg commented on GitHub (Aug 6, 2025):

Yeah, just fixed the actual command I used for installing (i.e. i installed into ./ollama not /usr)

I deleted the ./ollama dir before install and installed it again into the ./ollama. So I don't think that's the issue? I actually can't install it into /usr since i dont have rights there

<!-- gh-comment-id:3158065062 --> @timbmg commented on GitHub (Aug 6, 2025): Yeah, just fixed the actual command I used for installing (i.e. i installed into ./ollama not /usr) I deleted the ./ollama dir before install and installed it again into the ./ollama. So I don't think that's the issue? I actually can't install it into /usr since i dont have rights there
GiteaMirror commented 2026-05-04 19:28:52 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 6, 2025):

Is this a slurm job? Unset ROCR_VISIBLE_DEVICES in the start script.

<!-- gh-comment-id:3158344486 --> @rick-github commented on GitHub (Aug 6, 2025): Is this a slurm job? Unset `ROCR_VISIBLE_DEVICES` in the start script.
GiteaMirror commented 2026-05-04 19:28:52 -05:00
Author
Owner
Copy Link

@timbmg commented on GitHub (Aug 6, 2025):

Yes it's on Slurm.

Thanks, I tried that but the issue remains. I started ollama like this:
nvidia-smi & ROCR_VISIBLE_DEVICES="" OLLAMA_DEBUG=1 OLLAMA_LOAD_TIMEOUT=\"10m\" OLLAMA_HOST=0.0.0.0:${PORT} OLLAMA_MAX_LOADED_MODELS=1 ollama serve

which gave the log:
time=2025-08-06T11:10:00.461+02:00 level=INFO source=routes.go:1297 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0 GPU_DEVICE_ORDINAL:0 HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:DEBUG OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:10m0s OLLAMA_MAX_LOADED_MODELS:1 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/lab-storage-1/timbmg/.ollama/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[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: http_proxy: https_proxy: no_proxy:]"

<!-- gh-comment-id:3158493917 --> @timbmg commented on GitHub (Aug 6, 2025): Yes it's on Slurm. Thanks, I tried that but the issue remains. I started ollama like this: `nvidia-smi & ROCR_VISIBLE_DEVICES="" OLLAMA_DEBUG=1 OLLAMA_LOAD_TIMEOUT=\"10m\" OLLAMA_HOST=0.0.0.0:${PORT} OLLAMA_MAX_LOADED_MODELS=1 ollama serve` which gave the log: ` time=2025-08-06T11:10:00.461+02:00 level=INFO source=routes.go:1297 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0 GPU_DEVICE_ORDINAL:0 HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_CONTEXT_LENGTH:4096 OLLAMA_DEBUG:DEBUG OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:10m0s OLLAMA_MAX_LOADED_MODELS:1 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/lab-storage-1/timbmg/.ollama/models OLLAMA_MULTIUSER_CACHE:false OLLAMA_NEW_ENGINE:false OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[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: http_proxy: https_proxy: no_proxy:]" `
GiteaMirror commented 2026-05-04 19:28:53 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 6, 2025):

unset ROCR_VISIBLE_DEVICES
OLLAMA_DEBUG=1 OLLAMA_LOAD_TIMEOUT=10m OLLAMA_HOST=0.0.0.0:${PORT} OLLAMA_MAX_LOADED_MODELS=1 ollama serve
<!-- gh-comment-id:3158521846 --> @rick-github commented on GitHub (Aug 6, 2025): ``` unset ROCR_VISIBLE_DEVICES OLLAMA_DEBUG=1 OLLAMA_LOAD_TIMEOUT=10m OLLAMA_HOST=0.0.0.0:${PORT} OLLAMA_MAX_LOADED_MODELS=1 ollama serve ```
GiteaMirror commented 2026-05-04 19:28:54 -05:00
Author
Owner
Copy Link

@timbmg commented on GitHub (Aug 6, 2025):

Amazing, that worked! Thank you so much for your help 🤗

Any idea why I needed to unset that starting with the upgrade to 0.11?

<!-- gh-comment-id:3158557155 --> @timbmg commented on GitHub (Aug 6, 2025): Amazing, that worked! Thank you so much for your help 🤗 Any idea why I needed to unset that starting with the upgrade to 0.11?
GiteaMirror commented 2026-05-04 19:28:55 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 6, 2025):

https://github.com/ollama/ollama/pull/11169 introduced in 0.9.3 as a check for mixed libraries. Unclear as to why it didn't cause an issue for you earlier.

<!-- gh-comment-id:3158756089 --> @rick-github commented on GitHub (Aug 6, 2025): https://github.com/ollama/ollama/pull/11169 introduced in 0.9.3 as a check for mixed libraries. Unclear as to why it didn't cause an issue for you earlier.
GiteaMirror commented 2026-05-04 19:28:56 -05:00
Author
Owner
Copy Link

@Anxo06 commented on GitHub (Aug 12, 2025):

Solved the same issue for me! Thanks

<!-- gh-comment-id:3178929322 --> @Anxo06 commented on GitHub (Aug 12, 2025): Solved the same issue for me! Thanks
GiteaMirror commented 2026-05-04 19:28:57 -05:00
Author
Owner
Copy Link

@vuminhtue commented on GitHub (Sep 9, 2025):

Worked for me too. Thanks a lot @rick-github

<!-- gh-comment-id:3268877937 --> @vuminhtue commented on GitHub (Sep 9, 2025): Worked for me too. Thanks a lot @rick-github
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
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#69820
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?