github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #13726] Since version 0.13.2, ollama on arch cannot detect discret GPU Nvidia GeForce GTX 970M on hybrid laptop #55512

New Issue
Closed
opened 2026-04-29 09:19:22 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-29 09:19:22 -05:00
Owner
Copy Link

Originally created by @emiltoacs on GitHub (Jan 15, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/13726

Ollama stop detecting my Nvidia card since ollama 0.13.2 although it works with 0.13.1

I have an hybrid laptop (integrated GPU + discret GPU) with Arch Linux, running ollama-cuda12-bin because the Nvidia Maxwell cards
does not support CUDA 13. Since ollama version 0.13.2, ollama cannot run on my Nvidia GPU card
GTX 970M. Although in ollama-cuda12-bin version 0.13.1 my Nvidia card is well detected and
ollama runs on Nvidia.

Still buggy in version 0.14.1

nvidia-smi -L :

GPU 0: NVIDIA GeForce GTX 970M (UUID: GPU-ac90367f-1113-7f49-c527-d785b1684ce9)

Log when it works in 0.13.1

Log when ollama starts correctly in 0.13.1 :

source=routes.go:1544 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0,1,2,3 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:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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:/var/lib/ollama 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"
source=images.go:522 msg="total blobs: 26"
source=images.go:529 msg="total unused blobs removed: 0"
source=routes.go:1597 msg="Listening on 127.0.0.1:11434 (version 0.13.1)"
source=runner.go:67 msg="discovering available GPUs..."
source=runner.go:484 msg="user overrode visible devices" CUDA_VISIBLE_DEVICES=0,1,2,3
source=runner.go:488 msg="if GPUs are not correctly discovered, unset and try again"
source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 35545"
source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 37359"
source=types.go:42 msg="inference compute" id=GPU-ac90367f-1113-7f49-c527-d785b1684ce9 filter_id="" library=CUDA compute=5.2 name=CUDA0 description="NVIDIA GeForce GTX 970M" libdirs=ollama driver=13.0 pci_id=0000:01:00.0 type=discrete total="6.0 GiB" available="5.9 GiB"
source=routes.go:1638 msg="entering low vram mode" "total vram"="6.0 GiB" threshold="20.0 GiB"

Log when running model gemma3:4b-it-qat correctly in 0.13.1 for instance :

[GIN] 2026/01/15 - 09:54:29 | 200 |     129.625µs |       127.0.0.1 | HEAD     "/"
[GIN] 2026/01/15 - 09:54:29 | 200 |    4.826244ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2026/01/15 - 09:54:43 | 200 |      24.179µs |       127.0.0.1 | HEAD     "/"
[GIN] 2026/01/15 - 09:54:43 | 200 |  107.799622ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2026/01/15 - 09:54:43 | 200 |   84.906352ms |       127.0.0.1 | POST     "/api/show"
source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 42589"
source=server.go:209 msg="enabling flash attention"
source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /var/lib/ollama/blobs/sha256-529850705c0884a283b87d3b261d36ee30821e16f0310962ba977b456ad3b8cd --port 39057"
source=sched.go:443 msg="system memory" total="31.2 GiB" free="23.1 GiB" free_swap="47.5 GiB"
source=sched.go:450 msg="gpu memory" id=GPU-ac90367f-1113-7f49-c527-d785b1684ce9 library=CUDA available="5.4 GiB" free="5.9 GiB" minimum="457.0 MiB" overhead="0 B"
source=server.go:702 msg="loading model" "model layers"=35 requested=-1
source=runner.go:1398 msg="starting ollama engine"
source=runner.go:1433 msg="Server listening on 127.0.0.1:39057"
source=runner.go:1271 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:35[ID:GPU-ac90367f-1113-7f49-c527-d785b1684ce9 Layers:35(0..34)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
source=ggml.go:136 msg="" architecture=gemma3 file_type=Q4_0 name="Gemma3 4b It Qa_0 Qat Hf" description="" num_tensors=883 num_key_values=42
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 GTX 970M, compute capability 5.2, VMM: yes, ID: GPU-ac90367f-1113-7f49-c527-d785b1684ce9
load_backend: loaded CUDA backend from /usr/lib/ollama/libggml-cuda.so
load_backend: loaded CPU backend from /usr/lib/ollama/libggml-cpu-haswell.so
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 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(gcc)
source=runner.go:1271 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:35[ID:GPU-ac90367f-1113-7f49-c527-d785b1684ce9 Layers:35(0..34)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
source=runner.go:1271 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:35[ID:GPU-ac90367f-1113-7f49-c527-d785b1684ce9 Layers:35(0..34)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
source=ggml.go:482 msg="offloading 34 repeating layers to GPU"
source=ggml.go:489 msg="offloading output layer to GPU"
source=ggml.go:494 msg="offloaded 35/35 layers to GPU"
source=device.go:240 msg="model weights" device=CUDA0 size="3.7 GiB"
source=device.go:245 msg="model weights" device=CPU size="1.3 GiB"
source=device.go:251 msg="kv cache" device=CUDA0 size="254.0 MiB"
source=device.go:262 msg="compute graph" device=CUDA0 size="1.2 GiB"
source=device.go:267 msg="compute graph" device=CPU size="5.0 MiB"
source=device.go:272 msg="total memory" size="6.4 GiB"
source=sched.go:517 msg="loaded runners" count=1
source=server.go:1294 msg="waiting for llama runner to start responding"
source=server.go:1328 msg="waiting for server to become available" status="llm server loading model"
source=server.go:1332 msg="llama runner started in 9.58 seconds"
[GIN] 2026/01/15 - 09:54:53 | 200 |  9.961565254s |       127.0.0.1 | POST     "/api/generate"

Log when it fails in version 0.14.1

Log when ollama starts in 0.14.1 :

source=routes.go:1614 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0,1,2,3 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:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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:/var/lib/ollama 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"
source=images.go:499 msg="total blobs: 26"
source=images.go:506 msg="total unused blobs removed: 0"
source=routes.go:1667 msg="Listening on 127.0.0.1:11434 (version 0.14.1)"
source=runner.go:67 msg="discovering available GPUs..."
source=runner.go:485 msg="user overrode visible devices" CUDA_VISIBLE_DEVICES=0,1,2,3
source=runner.go:489 msg="if GPUs are not correctly discovered, unset and try again"
source=server.go:429 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 38909"
source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="31.2 GiB" available="17.3 GiB"
source=routes.go:1708 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB"
[GIN] 2026/01/15 - 10:17:05 | 200 |      70.962µs |       127.0.0.1 | GET      "/api/version"

Log when running model gemma3:4b-it-qat in 0.14.1 for instance :

[GIN] 2026/01/15 - 10:21:30 | 200 |      29.904µs |       127.0.0.1 | HEAD     "/"
[GIN] 2026/01/15 - 10:21:31 | 200 |  209.123423ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2026/01/15 - 10:21:31 | 200 |  189.735896ms |       127.0.0.1 | POST     "/api/show"
source=server.go:245 msg="enabling flash attention"
source=server.go:429 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /var/lib/ollama/blobs/sha256-529850705c0884a283b87d3b261d36ee30821e16f0310962ba977b456ad3b8cd --port 41595"
source=sched.go:452 msg="system memory" total="31.2 GiB" free="17.3 GiB" free_swap="47.3 GiB"
source=server.go:755 msg="loading model" "model layers"=35 requested=-1
source=runner.go:1405 msg="starting ollama engine"
source=runner.go:1440 msg="Server listening on 127.0.0.1:41595"
source=runner.go:1278 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:4 GPULayers:[] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
source=ggml.go:136 msg="" architecture=gemma3 file_type=Q4_0 name="Gemma3 4b It Qa_0 Qat Hf" description="" num_tensors=883 num_key_values=42
source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc)
source=runner.go:1278 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:4 GPULayers:[] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
source=runner.go:1278 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:4 GPULayers:[] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}"
source=ggml.go:482 msg="offloading 0 repeating layers to GPU"
source=ggml.go:486 msg="offloading output layer to CPU"
source=ggml.go:494 msg="offloaded 0/35 layers to GPU"
source=device.go:245 msg="model weights" device=CPU size="5.0 GiB"
source=device.go:256 msg="kv cache" device=CPU size="254.0 MiB"
source=device.go:267 msg="compute graph" device=CPU size="126.9 MiB"
source=device.go:272 msg="total memory" size="5.3 GiB"
source=sched.go:526 msg="loaded runners" count=1
source=server.go:1347 msg="waiting for llama runner to start responding"
source=server.go:1381 msg="waiting for server to become available" status="llm server loading model"
source=server.go:1385 msg="llama runner started in 8.88 seconds"
[GIN] 2026/01/15 - 10:21:40 | 200 |  9.166107971s |       127.0.0.1 | POST     "/api/generate"

OS

Linux 6.18.5-arch1-1

GPU

Intel Corporation Skylake-H GT2 [HD Graphics 530]
NVIDIA Corporation GM204M [GeForce GTX 960 OEM / 970M]

CPU

Intel Core i7-6700HQ

Ollama version

0.14.1

Originally created by @emiltoacs on GitHub (Jan 15, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/13726 # Ollama stop detecting my Nvidia card since ollama 0.13.2 although it works with 0.13.1 I have an hybrid laptop (integrated GPU + discret GPU) with Arch Linux, running `ollama-cuda12-bin` because the Nvidia Maxwell cards does not support CUDA 13. Since ollama version `0.13.2`, ollama cannot run on my Nvidia GPU card GTX 970M. Although in `ollama-cuda12-bin` version `0.13.1` my Nvidia card is well detected and ollama runs on Nvidia. Still buggy in version `0.14.1` nvidia-smi -L : GPU 0: NVIDIA GeForce GTX 970M (UUID: GPU-ac90367f-1113-7f49-c527-d785b1684ce9) <a id="orgc4c4d42"></a> # Log when it works in 0.13.1 Log when ollama starts correctly in 0.13.1 : source=routes.go:1544 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0,1,2,3 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:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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:/var/lib/ollama 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]" source=images.go:522 msg="total blobs: 26" source=images.go:529 msg="total unused blobs removed: 0" source=routes.go:1597 msg="Listening on 127.0.0.1:11434 (version 0.13.1)" source=runner.go:67 msg="discovering available GPUs..." source=runner.go:484 msg="user overrode visible devices" CUDA_VISIBLE_DEVICES=0,1,2,3 source=runner.go:488 msg="if GPUs are not correctly discovered, unset and try again" source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 35545" source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 37359" source=types.go:42 msg="inference compute" id=GPU-ac90367f-1113-7f49-c527-d785b1684ce9 filter_id="" library=CUDA compute=5.2 name=CUDA0 description="NVIDIA GeForce GTX 970M" libdirs=ollama driver=13.0 pci_id=0000:01:00.0 type=discrete total="6.0 GiB" available="5.9 GiB" source=routes.go:1638 msg="entering low vram mode" "total vram"="6.0 GiB" threshold="20.0 GiB" Log when running model gemma3:4b-it-qat correctly in 0.13.1 for instance : [GIN] 2026/01/15 - 09:54:29 | 200 | 129.625µs | 127.0.0.1 | HEAD "/" [GIN] 2026/01/15 - 09:54:29 | 200 | 4.826244ms | 127.0.0.1 | GET "/api/tags" [GIN] 2026/01/15 - 09:54:43 | 200 | 24.179µs | 127.0.0.1 | HEAD "/" [GIN] 2026/01/15 - 09:54:43 | 200 | 107.799622ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/01/15 - 09:54:43 | 200 | 84.906352ms | 127.0.0.1 | POST "/api/show" source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 42589" source=server.go:209 msg="enabling flash attention" source=server.go:392 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /var/lib/ollama/blobs/sha256-529850705c0884a283b87d3b261d36ee30821e16f0310962ba977b456ad3b8cd --port 39057" source=sched.go:443 msg="system memory" total="31.2 GiB" free="23.1 GiB" free_swap="47.5 GiB" source=sched.go:450 msg="gpu memory" id=GPU-ac90367f-1113-7f49-c527-d785b1684ce9 library=CUDA available="5.4 GiB" free="5.9 GiB" minimum="457.0 MiB" overhead="0 B" source=server.go:702 msg="loading model" "model layers"=35 requested=-1 source=runner.go:1398 msg="starting ollama engine" source=runner.go:1433 msg="Server listening on 127.0.0.1:39057" source=runner.go:1271 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:35[ID:GPU-ac90367f-1113-7f49-c527-d785b1684ce9 Layers:35(0..34)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" source=ggml.go:136 msg="" architecture=gemma3 file_type=Q4_0 name="Gemma3 4b It Qa_0 Qat Hf" description="" num_tensors=883 num_key_values=42 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 GTX 970M, compute capability 5.2, VMM: yes, ID: GPU-ac90367f-1113-7f49-c527-d785b1684ce9 load_backend: loaded CUDA backend from /usr/lib/ollama/libggml-cuda.so load_backend: loaded CPU backend from /usr/lib/ollama/libggml-cpu-haswell.so 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 CUDA.0.ARCHS=500,520,600,610,700,750,800,860,890,900,1200 CUDA.0.USE_GRAPHS=1 CUDA.0.PEER_MAX_BATCH_SIZE=128 compiler=cgo(gcc) source=runner.go:1271 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:35[ID:GPU-ac90367f-1113-7f49-c527-d785b1684ce9 Layers:35(0..34)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" source=runner.go:1271 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:true KvSize:4096 KvCacheType: NumThreads:4 GPULayers:35[ID:GPU-ac90367f-1113-7f49-c527-d785b1684ce9 Layers:35(0..34)] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" source=ggml.go:482 msg="offloading 34 repeating layers to GPU" source=ggml.go:489 msg="offloading output layer to GPU" source=ggml.go:494 msg="offloaded 35/35 layers to GPU" source=device.go:240 msg="model weights" device=CUDA0 size="3.7 GiB" source=device.go:245 msg="model weights" device=CPU size="1.3 GiB" source=device.go:251 msg="kv cache" device=CUDA0 size="254.0 MiB" source=device.go:262 msg="compute graph" device=CUDA0 size="1.2 GiB" source=device.go:267 msg="compute graph" device=CPU size="5.0 MiB" source=device.go:272 msg="total memory" size="6.4 GiB" source=sched.go:517 msg="loaded runners" count=1 source=server.go:1294 msg="waiting for llama runner to start responding" source=server.go:1328 msg="waiting for server to become available" status="llm server loading model" source=server.go:1332 msg="llama runner started in 9.58 seconds" [GIN] 2026/01/15 - 09:54:53 | 200 | 9.961565254s | 127.0.0.1 | POST "/api/generate" <a id="orga7fee46"></a> # Log when it fails in version 0.14.1 Log when ollama starts in 0.14.1 : source=routes.go:1614 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0,1,2,3 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:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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:/var/lib/ollama 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]" source=images.go:499 msg="total blobs: 26" source=images.go:506 msg="total unused blobs removed: 0" source=routes.go:1667 msg="Listening on 127.0.0.1:11434 (version 0.14.1)" source=runner.go:67 msg="discovering available GPUs..." source=runner.go:485 msg="user overrode visible devices" CUDA_VISIBLE_DEVICES=0,1,2,3 source=runner.go:489 msg="if GPUs are not correctly discovered, unset and try again" source=server.go:429 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 38909" source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="31.2 GiB" available="17.3 GiB" source=routes.go:1708 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB" [GIN] 2026/01/15 - 10:17:05 | 200 | 70.962µs | 127.0.0.1 | GET "/api/version" Log when running model gemma3:4b-it-qat in 0.14.1 for instance : [GIN] 2026/01/15 - 10:21:30 | 200 | 29.904µs | 127.0.0.1 | HEAD "/" [GIN] 2026/01/15 - 10:21:31 | 200 | 209.123423ms | 127.0.0.1 | POST "/api/show" [GIN] 2026/01/15 - 10:21:31 | 200 | 189.735896ms | 127.0.0.1 | POST "/api/show" source=server.go:245 msg="enabling flash attention" source=server.go:429 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --model /var/lib/ollama/blobs/sha256-529850705c0884a283b87d3b261d36ee30821e16f0310962ba977b456ad3b8cd --port 41595" source=sched.go:452 msg="system memory" total="31.2 GiB" free="17.3 GiB" free_swap="47.3 GiB" source=server.go:755 msg="loading model" "model layers"=35 requested=-1 source=runner.go:1405 msg="starting ollama engine" source=runner.go:1440 msg="Server listening on 127.0.0.1:41595" source=runner.go:1278 msg=load request="{Operation:fit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:4 GPULayers:[] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" source=ggml.go:136 msg="" architecture=gemma3 file_type=Q4_0 name="Gemma3 4b It Qa_0 Qat Hf" description="" num_tensors=883 num_key_values=42 source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc) source=runner.go:1278 msg=load request="{Operation:alloc LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:4 GPULayers:[] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" source=runner.go:1278 msg=load request="{Operation:commit LoraPath:[] Parallel:1 BatchSize:512 FlashAttention:Enabled KvSize:4096 KvCacheType: NumThreads:4 GPULayers:[] MultiUserCache:false ProjectorPath: MainGPU:0 UseMmap:false}" source=ggml.go:482 msg="offloading 0 repeating layers to GPU" source=ggml.go:486 msg="offloading output layer to CPU" source=ggml.go:494 msg="offloaded 0/35 layers to GPU" source=device.go:245 msg="model weights" device=CPU size="5.0 GiB" source=device.go:256 msg="kv cache" device=CPU size="254.0 MiB" source=device.go:267 msg="compute graph" device=CPU size="126.9 MiB" source=device.go:272 msg="total memory" size="5.3 GiB" source=sched.go:526 msg="loaded runners" count=1 source=server.go:1347 msg="waiting for llama runner to start responding" source=server.go:1381 msg="waiting for server to become available" status="llm server loading model" source=server.go:1385 msg="llama runner started in 8.88 seconds" [GIN] 2026/01/15 - 10:21:40 | 200 | 9.166107971s | 127.0.0.1 | POST "/api/generate" ### OS Linux 6.18.5-arch1-1 ### GPU Intel Corporation Skylake-H GT2 [HD Graphics 530] NVIDIA Corporation GM204M [GeForce GTX 960 OEM / 970M] ### CPU Intel Core i7-6700HQ ### Ollama version 0.14.1
GiteaMirror added the bug label 2026-04-29 09:19:22 -05:00
GiteaMirror commented 2026-04-29 09:19:27 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 15, 2026):

Set OLLAMA_DEBUG=2 in the server environment and post the log.

Why have you set CUDA_VISIBLE_DEVICES?

<!-- gh-comment-id:3753831839 --> @rick-github commented on GitHub (Jan 15, 2026): Set `OLLAMA_DEBUG=2` in the server environment and post the log. Why have you set `CUDA_VISIBLE_DEVICES`?
GiteaMirror commented 2026-04-29 09:19:28 -05:00
Author
Owner
Copy Link

@emiltoacs commented on GitHub (Jan 15, 2026):

Ooops I tried different value with CUDA_VISIBLE_DEVICES and forgot to remove it. I unset the variable.

There is the log with OLLAMA_DEBUG=2

source=routes.go:1614 msg="server config" env="map[CUDA_VISIBLE_DEVICES: 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:DEBUG-4 OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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:/var/lib/ollama 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"
source=images.go:499 msg="total blobs: 26"
source=images.go:506 msg="total unused blobs removed: 0"
source=routes.go:1667 msg="Listening on 127.0.0.1:11434 (version 0.14.1)"
 source=sched.go:121 msg="starting llm scheduler"
source=runner.go:67 msg="discovering available GPUs..."
 source=runner.go:440 msg="starting runner for device discovery" libDirs=[/usr/lib/ollama] extraEnvs=map[]
source=server.go:429 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 34747"
 source=server.go:430 msg=subprocess PATH=/usr/local/sbin:/usr/local/bin:/usr/bin OLLAMA_MODELS=/var/lib/ollama CUDA_DEVICE_ORDER=PCI_BUS_ID CUDA_LAUNCH_BLOCKING=0 CUDA_MODULE_LOADING=LAZY CUDA_MODULE_DATA_LOADING=LAZY CUDA_CACHE_MAXSIZE=2147483648 CUDA_CACHE_PATH=/var/cache/cuda CUDA_LOG_FILE=/var/log/cuda.log OLLAMA_HOST=http://127.0.0.1:11434 OLLAMA_KEEP_ALIVE=5m OLLAMA_LOAD_TIMEOUT=5m OLLAMA_CONTEXT_LENGTH=4096 OLLAMA_MAX_LOADED_MODELS=0 OLLAMA_GPU_OVERHEAD=0 OLLAMA_MAX_QUEUE=512 OLLAMA_NUM_PARALLEL=1 OLLAMA_NOHISTORY=false OLLAMA_NOPRUNE=false OLLAMA_FLASH_ATTENTION=true OLLAMA_SCHED_SPREAD=false OLLAMA_DEBUG=2 LD_LIBRARY_PATH=/usr/lib/ollama OLLAMA_LIBRARY_PATH=/usr/lib/ollama
source=runner.go:1405 msg="starting ollama engine"
source=runner.go:1440 msg="Server listening on 127.0.0.1:34747"
 source=gguf.go:589 msg=general.architecture type=string
 source=gguf.go:589 msg=tokenizer.ggml.model type=string
 source=ggml.go:296 msg="key with type not found" key=general.alignment default=32
 source=ggml.go:296 msg="key with type not found" key=general.alignment default=32
 source=ggml.go:296 msg="key with type not found" key=general.file_type default=0
 source=ggml.go:296 msg="key with type not found" key=general.name default=""
 source=ggml.go:296 msg="key with type not found" key=general.description default=""
source=ggml.go:136 msg="" architecture=llama file_type=unknown name="" description="" num_tensors=0 num_key_values=3
 source=ggml.go:94 msg="ggml backend load all from path" path=/usr/lib/ollama
source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc)
 source=ggml.go:296 msg="key with type not found" key=llama.block_count default=0
 source=ggml.go:296 msg="key with type not found" key=llama.pooling_type default=0
 source=ggml.go:296 msg="key with type not found" key=llama.expert_count default=0
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.tokens default="&{size:0 values:[]}"
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.scores default="&{size:0 values:[]}"
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.token_type default="&{size:0 values:[]}"
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.merges default="&{size:0 values:[]}"
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.add_bos_token default=true
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.bos_token_id default=0
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.add_eos_token default=false
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.eos_token_id default=0
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.eos_token_ids default="&{size:0 values:[]}"
 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.pre default=""
 source=ggml.go:296 msg="key with type not found" key=llama.block_count default=0
 source=ggml.go:296 msg="key with type not found" key=llama.embedding_length default=0
 source=ggml.go:296 msg="key with type not found" key=llama.attention.head_count default=0
 source=ggml.go:296 msg="key with type not found" key=llama.attention.head_count_kv default=0
 source=ggml.go:296 msg="key with type not found" key=llama.attention.key_length default=0
 source=ggml.go:296 msg="key with type not found" key=llama.rope.dimension_count default=0
 source=ggml.go:296 msg="key with type not found" key=llama.attention.layer_norm_rms_epsilon default=0
 source=ggml.go:296 msg="key with type not found" key=llama.rope.freq_base default=100000
 source=ggml.go:296 msg="key with type not found" key=llama.rope.scaling.factor default=1
 source=runner.go:1380 msg="dummy model load took" duration=7.001259ms
 source=runner.go:1385 msg="gathering device infos took" duration=711ns
 source=runner.go:467 msg="runner enumerated devices" OLLAMA_LIBRARY_PATH=[/usr/lib/ollama] devices=[]
 source=runner.go:437 msg="bootstrap discovery took" duration=30.816301ms OLLAMA_LIBRARY_PATH=[/usr/lib/ollama] extra_envs=map[]
 source=runner.go:124 msg="evaluating which, if any, devices to filter out" initial_count=0
 source=runner.go:174 msg="supported GPU library combinations before filtering" supported=map[]
 source=runner.go:40 msg="GPU bootstrap discovery took" duration=31.149958ms
source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="31.2 GiB" available="14.1 GiB"
source=routes.go:1708 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB"
<!-- gh-comment-id:3753989858 --> @emiltoacs commented on GitHub (Jan 15, 2026): Ooops I tried different value with CUDA_VISIBLE_DEVICES and forgot to remove it. I unset the variable. There is the log with OLLAMA_DEBUG=2 source=routes.go:1614 msg="server config" env="map[CUDA_VISIBLE_DEVICES: 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:DEBUG-4 OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 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:/var/lib/ollama 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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]" source=images.go:499 msg="total blobs: 26" source=images.go:506 msg="total unused blobs removed: 0" source=routes.go:1667 msg="Listening on 127.0.0.1:11434 (version 0.14.1)" source=sched.go:121 msg="starting llm scheduler" source=runner.go:67 msg="discovering available GPUs..." source=runner.go:440 msg="starting runner for device discovery" libDirs=[/usr/lib/ollama] extraEnvs=map[] source=server.go:429 msg="starting runner" cmd="/usr/bin/ollama runner --ollama-engine --port 34747" source=server.go:430 msg=subprocess PATH=/usr/local/sbin:/usr/local/bin:/usr/bin OLLAMA_MODELS=/var/lib/ollama CUDA_DEVICE_ORDER=PCI_BUS_ID CUDA_LAUNCH_BLOCKING=0 CUDA_MODULE_LOADING=LAZY CUDA_MODULE_DATA_LOADING=LAZY CUDA_CACHE_MAXSIZE=2147483648 CUDA_CACHE_PATH=/var/cache/cuda CUDA_LOG_FILE=/var/log/cuda.log OLLAMA_HOST=http://127.0.0.1:11434 OLLAMA_KEEP_ALIVE=5m OLLAMA_LOAD_TIMEOUT=5m OLLAMA_CONTEXT_LENGTH=4096 OLLAMA_MAX_LOADED_MODELS=0 OLLAMA_GPU_OVERHEAD=0 OLLAMA_MAX_QUEUE=512 OLLAMA_NUM_PARALLEL=1 OLLAMA_NOHISTORY=false OLLAMA_NOPRUNE=false OLLAMA_FLASH_ATTENTION=true OLLAMA_SCHED_SPREAD=false OLLAMA_DEBUG=2 LD_LIBRARY_PATH=/usr/lib/ollama OLLAMA_LIBRARY_PATH=/usr/lib/ollama source=runner.go:1405 msg="starting ollama engine" source=runner.go:1440 msg="Server listening on 127.0.0.1:34747" source=gguf.go:589 msg=general.architecture type=string source=gguf.go:589 msg=tokenizer.ggml.model type=string source=ggml.go:296 msg="key with type not found" key=general.alignment default=32 source=ggml.go:296 msg="key with type not found" key=general.alignment default=32 source=ggml.go:296 msg="key with type not found" key=general.file_type default=0 source=ggml.go:296 msg="key with type not found" key=general.name default="" source=ggml.go:296 msg="key with type not found" key=general.description default="" source=ggml.go:136 msg="" architecture=llama file_type=unknown name="" description="" num_tensors=0 num_key_values=3 source=ggml.go:94 msg="ggml backend load all from path" path=/usr/lib/ollama source=ggml.go:104 msg=system CPU.0.LLAMAFILE=1 compiler=cgo(gcc) source=ggml.go:296 msg="key with type not found" key=llama.block_count default=0 source=ggml.go:296 msg="key with type not found" key=llama.pooling_type default=0 source=ggml.go:296 msg="key with type not found" key=llama.expert_count default=0 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.tokens default="&{size:0 values:[]}" source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.scores default="&{size:0 values:[]}" source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.token_type default="&{size:0 values:[]}" source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.merges default="&{size:0 values:[]}" source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.add_bos_token default=true source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.bos_token_id default=0 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.add_eos_token default=false source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.eos_token_id default=0 source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.eos_token_ids default="&{size:0 values:[]}" source=ggml.go:296 msg="key with type not found" key=tokenizer.ggml.pre default="" source=ggml.go:296 msg="key with type not found" key=llama.block_count default=0 source=ggml.go:296 msg="key with type not found" key=llama.embedding_length default=0 source=ggml.go:296 msg="key with type not found" key=llama.attention.head_count default=0 source=ggml.go:296 msg="key with type not found" key=llama.attention.head_count_kv default=0 source=ggml.go:296 msg="key with type not found" key=llama.attention.key_length default=0 source=ggml.go:296 msg="key with type not found" key=llama.rope.dimension_count default=0 source=ggml.go:296 msg="key with type not found" key=llama.attention.layer_norm_rms_epsilon default=0 source=ggml.go:296 msg="key with type not found" key=llama.rope.freq_base default=100000 source=ggml.go:296 msg="key with type not found" key=llama.rope.scaling.factor default=1 source=runner.go:1380 msg="dummy model load took" duration=7.001259ms source=runner.go:1385 msg="gathering device infos took" duration=711ns source=runner.go:467 msg="runner enumerated devices" OLLAMA_LIBRARY_PATH=[/usr/lib/ollama] devices=[] source=runner.go:437 msg="bootstrap discovery took" duration=30.816301ms OLLAMA_LIBRARY_PATH=[/usr/lib/ollama] extra_envs=map[] source=runner.go:124 msg="evaluating which, if any, devices to filter out" initial_count=0 source=runner.go:174 msg="supported GPU library combinations before filtering" supported=map[] source=runner.go:40 msg="GPU bootstrap discovery took" duration=31.149958ms source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="31.2 GiB" available="14.1 GiB" source=routes.go:1708 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB"
GiteaMirror commented 2026-04-29 09:19:29 -05:00
Author
Owner
Copy Link

@emiltoacs commented on GitHub (Jan 26, 2026):

The bug seems to affect only the aur package ollama-cuda12-bin, when you install ollama
version manually with the archive from the ollama website :
https://ollama.com/download/ollama-linux-amd64.tar.zst
at least from version 14.1 it works.

<!-- gh-comment-id:3801362330 --> @emiltoacs commented on GitHub (Jan 26, 2026): The bug seems to affect only the aur package `ollama-cuda12-bin`, when you install ollama version manually with the archive from the ollama website : <https://ollama.com/download/ollama-linux-amd64.tar.zst> at least from version 14.1 it works.
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#55512
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?