github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 13:19:40 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #13593] Ollama does not detect GPU (RTX A1000 Laptop, Windows 11, WDDM) even after driver update #55463

New Issue
Open
opened 2026-04-29 09:15:56 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-29 09:15:56 -05:00
Owner
Copy Link

Originally created by @hrithnajmi on GitHub (Dec 31, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13593

What is the issue?

Describe the issue

Ollama is not detecting or using my NVIDIA GPU for inference on Windows 11, even after updating to the latest drivers. It always runs in CPU mode and reports "total vram=0 B" and "inference compute id=cpu" in the logs.

System Info

  • Ollama version: 0.13.5
  • Windows version: Windows 11 64-bit
  • GPU: NVIDIA RTX A1000 Laptop GPU
  • NVIDIA driver version: 582.08
  • CUDA version (from nvidia-smi): 13.0
  • Laptop/Hybrid graphics: Yes (WDDM mode)
  • Model tested: mistral-nemo (GPU-accelerated model)

Steps to reproduce

  1. Install latest NVIDIA drivers (clean install, rebooted)
  2. Set NVIDIA Control Panel to use "High-performance NVIDIA processor" for ollama.exe
  3. Run ollama serve from command prompt (as admin)
  4. Send a prompt to a GPU-supported model
  5. Observe logs and check nvidia-smi (no Ollama process uses GPU, logs show CPU only)

Additional context

  • This same device was able to use the GPU with Ollama earlier in the day, but now only CPU is detected.

Relevant log output

time=... level=INFO source=types.go:60 msg="inference compute" id=cpu library=cpu ...
time=... level=INFO source=routes.go:1648 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB"

OS

Windows

GPU

Nvidia

CPU

Intel

Ollama version

0.13.5

Originally created by @hrithnajmi on GitHub (Dec 31, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13593 ### What is the issue? ### Describe the issue Ollama is not detecting or using my NVIDIA GPU for inference on Windows 11, even after updating to the latest drivers. It always runs in CPU mode and reports "total vram=0 B" and "inference compute id=cpu" in the logs. ### System Info - Ollama version: 0.13.5 - Windows version: Windows 11 64-bit - GPU: NVIDIA RTX A1000 Laptop GPU - NVIDIA driver version: 582.08 - CUDA version (from nvidia-smi): 13.0 - Laptop/Hybrid graphics: Yes (WDDM mode) - Model tested: mistral-nemo (GPU-accelerated model) ### Steps to reproduce 1. Install latest NVIDIA drivers (clean install, rebooted) 2. Set NVIDIA Control Panel to use "High-performance NVIDIA processor" for ollama.exe 3. Run `ollama serve` from command prompt (as admin) 4. Send a prompt to a GPU-supported model 5. Observe logs and check `nvidia-smi` (no Ollama process uses GPU, logs show CPU only) ### Additional context - This same device was able to use the GPU with Ollama earlier in the day, but now only CPU is detected. ### Relevant log output ```shell time=... level=INFO source=types.go:60 msg="inference compute" id=cpu library=cpu ... time=... level=INFO source=routes.go:1648 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB" ``` ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version 0.13.5
GiteaMirror added the bug label 2026-04-29 09:15:56 -05:00
GiteaMirror commented 2026-04-29 09:16:00 -05:00
Author
Owner
Copy Link

@D337z commented on GitHub (Dec 31, 2025):

What laptop brand is this (UEFI can affect the GPU)? Does it still not see the GPU after restarting Ollama? And is your screen going into low-power state (blank screen) in between drop-outs?

<!-- gh-comment-id:3702720617 --> @D337z commented on GitHub (Dec 31, 2025): What laptop brand is this (UEFI can affect the GPU)? Does it still not see the GPU after restarting Ollama? And is your screen going into low-power state (blank screen) in between drop-outs?
GiteaMirror commented 2026-04-29 09:16:00 -05:00
Author
Owner
Copy Link

@hrithnajmi commented on GitHub (Dec 31, 2025):

What laptop brand is this (UEFI can affect the GPU)? Does it still not see the GPU after restarting Ollama? And is your screen going into low-power state (blank screen) in between drop-outs?

Hi, I'm using a Dell Precision 5570. And Yes, I have restarted Ollama multiple times, rebooting the laptop and even installing the latest NVIDIA drivers. But the issue still persists, Ollama still does not detect the GPU and runs in CPU mode. No, my screen does not go into a low-power state or blank out in between drop-outs. The display remains normal during all attempts.

Additional Info:

  • NVIDIA RTX A1000 Laptop GPU
  • Driver version: 582.08
  • CUDA version: 13.0
  • Ollama version: 0.13.5
  • “High-performance NVIDIA processor” is set for ollama.exe in NVIDIA Control Panel
  • nvidia-smi shows the GPU is present and idle, but Ollama logs always show “inference compute id=cpu” and “total vram=0 B”
<!-- gh-comment-id:3702735361 --> @hrithnajmi commented on GitHub (Dec 31, 2025): > What laptop brand is this (UEFI can affect the GPU)? Does it still not see the GPU after restarting Ollama? And is your screen going into low-power state (blank screen) in between drop-outs? Hi, I'm using a Dell Precision 5570. And Yes, I have restarted Ollama multiple times, rebooting the laptop and even installing the latest NVIDIA drivers. But the issue still persists, Ollama still does not detect the GPU and runs in CPU mode. No, my screen does not go into a low-power state or blank out in between drop-outs. The display remains normal during all attempts. ### Additional Info: - NVIDIA RTX A1000 Laptop GPU - Driver version: 582.08 - CUDA version: 13.0 - Ollama version: 0.13.5 - “High-performance NVIDIA processor” is set for ollama.exe in NVIDIA Control Panel - nvidia-smi shows the GPU is present and idle, but Ollama logs always show “inference compute id=cpu” and “total vram=0 B”
GiteaMirror commented 2026-04-29 09:16:00 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Dec 31, 2025):

Set OLLAMA_DEBUG=2 and post the log from start to inference compute.

<!-- gh-comment-id:3702890783 --> @rick-github commented on GitHub (Dec 31, 2025): Set `OLLAMA_DEBUG=2` and post the log from start to `inference compute`.
GiteaMirror commented 2026-04-29 09:16:02 -05:00
Author
Owner
Copy Link

@hrithnajmi commented on GitHub (Jan 1, 2026):

Set OLLAMA_DEBUG=2 and post the log from start to inference compute.

time=2026-01-01T09:40:01.222+08:00 level=INFO source=routes.go:1554 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:false 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:D:\\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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES:]"
time=2026-01-01T09:40:01.239+08:00 level=INFO source=images.go:493 msg="total blobs: 38"
time=2026-01-01T09:40:01.240+08:00 level=INFO source=images.go:500 msg="total unused blobs removed: 0"
time=2026-01-01T09:40:01.241+08:00 level=INFO source=routes.go:1607 msg="Listening on 127.0.0.1:11434 (version 0.13.5)"
time=2026-01-01T09:40:01.288+08:00 level=DEBUG source=sched.go:120 msg="starting llm scheduler"
time=2026-01-01T09:40:01.289+08:00 level=INFO source=runner.go:67 msg="discovering available GPUs..."
time=2026-01-01T09:40:01.297+08:00 level=TRACE source=runner.go:440 msg="starting runner for device discovery" libDirs=[D:\ollama\lib\ollama] extraEnvs=map[]
time=2026-01-01T09:40:01.304+08:00 level=INFO source=server.go:429 msg="starting runner" cmd="D:\\ollama\\ollama.exe runner --ollama-engine --port 51152"
time=2026-01-01T09:40:01.304+08:00 level=DEBUG source=server.go:430 msg=subprocess CUDA_PATH="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8" CUDA_PATH_V11_6="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8" OLLAMA_DEBUG=2 OLLAMA_MODELS=D:\ollama\models OLLAMA_USE_GPU=1 PATH="D:\\ollama\\lib\\ollama;D:\\Sankyu GitHub\\sankyu chatbot\\Sankyu_Academy_BE\\venv\\Scripts;c:\\Users\\pierr\\AppData\\Roaming\\Code\\User\\globalStorage\\github.copilot-chat\\debugCommand;c:\\Users\\pierr\\AppData\\Roaming\\Code\\User\\globalStorage\\github.copilot-chat\\copilotCli;C:\\Program Files\\Basler\\pylon 7\\Runtime\\x64\\;C:\\Program Files\\Basler\\pylon 7\\Runtime\\Win32\\;C:\\Program Files (x86)\\Common Files\\MVS\\Runtime\\Win32_i86;C:\\Program Files (x86)\\Common Files\\MVS\\Runtime\\Win64_x64;C:\\Python312\\Scripts\\;C:\\Python312\\;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\Scripts\\;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\bin;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\libnvvp;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\extras\\CUPTI\\lib64;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\extras\\CUPTI\\include;C:\\WINDOWS\\system32\\config\\systemprofile\\AppData\\Local\\Microsoft\\WindowsApps;C:\\Users\\pierr\\AppData\\Roaming\\npm;C:\\WINDOWS\\system32\\config\\systemprofile\\.dotnet\\tools;C:\\Users\\pierr\\AppData\\Local\\Programs\\Microsoft VS Code\\bin;C:\\Users\\pierr\\AppData\\Local\\GitHubDesktop\\bin;C:\\Program Files (x86)\\poppler-24.08.0\\Library\\bin;C:\\Program Files\\Git\\bin;C:\\Program Files\\dotnet\\;C:\\poppler-24.08.0\\Library\\bin;C:\\Program Files\\nodejs\\;C:\\ProgramData\\chocolatey\\bin;C:\\Windows\\System32\\nvidia-smi.exe;C:\\Program Files\\NVIDIA Corporation\\NVSMI;C:\\WINDOWS\\system32;C:\\WINDOWS;C:\\WINDOWS\\System32\\Wbem;C:\\WINDOWS\\System32\\WindowsPowerShell\\v1.0\\;C:\\WINDOWS\\System32\\OpenSSH\\;C:\\Program Files\\NVIDIA Corporation\\NVIDIA app\\NvDLISR;;C:\\Program Files\\Docker\\Docker\\resources\\bin;C:\\Users\\pierr\\.local\\bin;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\Scripts\\;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\bin;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\libnvvp;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\extras\\CUPTI\\lib64;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\extras\\CUPTI\\include;C:\\Users\\pierr\\AppData\\Local\\Microsoft\\WindowsApps;C:\\Users\\pierr\\.dotnet\\tools;C:\\Users\\pierr\\AppData\\Local\\Programs\\Microsoft VS Code\\bin;C:\\Users\\pierr\\AppData\\Local\\GitHubDesktop\\bin;C:\\Users\\pierr\\AppData\\Roaming\\npm;C:\\Users\\pierr\\AppData\\Local\\Programs\\cursor\\resources\\app\\bin;C:\\Users\\pierr\\AppData\\Local\\Programs\\Ollama;D:\\ollama;C:\\Users\\pierr\\Downloads\\ffmpeg-2025-08-11-git-3542260376-full_build\\ffmpeg-2025-08-11-git-3542260376-full_build\\bin;C:\\Users\\pierr\\Downloads\\Release-24.08.0-0 (2)\\Release-24.08.0-0\\poppler-24.08.0\\Library\\bin;c:\\Users\\pierr\\.vscode\\extensions\\ms-python.debugpy-2025.18.0-win32-x64\\bundled\\scripts\\noConfigScripts" OLLAMA_LIBRARY_PATH=D:\ollama\lib\ollama
time=2026-01-01T09:40:01.340+08:00 level=INFO source=runner.go:1405 msg="starting ollama engine"
time=2026-01-01T09:40:01.353+08:00 level=INFO source=runner.go:1440 msg="Server listening on 127.0.0.1:51152"
time=2026-01-01T09:40:01.358+08:00 level=DEBUG source=gguf.go:589 msg=general.architecture type=string
time=2026-01-01T09:40:01.359+08:00 level=DEBUG source=gguf.go:589 msg=tokenizer.ggml.model type=string
time=2026-01-01T09:40:01.359+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.alignment default=32
time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.alignment default=32
time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.file_type default=0
time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.name default=""
time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.description default=""
time=2026-01-01T09:40:01.360+08:00 level=INFO source=ggml.go:136 msg="" architecture=llama file_type=unknown name="" description="" num_tensors=0 num_key_values=3  
time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:94 msg="ggml backend load all from path" path=D:\ollama\lib\ollama
load_backend: loaded CPU backend from D:\ollama\lib\ollama\ggml-cpu-alderlake.dll
time=2026-01-01T09:40:01.374+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX_VNNI=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(clang)
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.block_count default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.pooling_type default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.expert_count default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.tokens default="&{size:0 values:[]}"
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.scores default="&{size:0 values:[]}"
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.token_type default="&{size:0 values:[]}"
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.merges default="&{size:0 values:[]}"
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.add_bos_token default=true
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.bos_token_id default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.add_eos_token default=false
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.eos_token_id default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.eos_token_ids default="&{size:0 values:[]}"      
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.pre default=""
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.block_count default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.embedding_length default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.head_count default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.head_count_kv default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.key_length default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.rope.dimension_count default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.layer_norm_rms_epsilon default=0
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.rope.freq_base default=100000
time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.rope.scaling.factor default=1
time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:1380 msg="dummy model load took" duration=16.6606ms
time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:1385 msg="gathering device infos took" duration=0s
time=2026-01-01T09:40:01.375+08:00 level=TRACE source=runner.go:467 msg="runner enumerated devices" OLLAMA_LIBRARY_PATH=[D:\ollama\lib\ollama] devices=[]
time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:437 msg="bootstrap discovery took" duration=78.2778ms OLLAMA_LIBRARY_PATH=[D:\ollama\lib\ollama] extra_envs=map[]
time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:124 msg="evaluating which, if any, devices to filter out" initial_count=0
time=2026-01-01T09:40:01.375+08:00 level=TRACE source=runner.go:174 msg="supported GPU library combinations before filtering" supported=map[]
time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:40 msg="GPU bootstrap discovery took" duration=87.247ms
time=2026-01-01T09:40:01.376+08:00 level=INFO source=types.go:60 msg="inference compute"
<!-- gh-comment-id:3703151942 --> @hrithnajmi commented on GitHub (Jan 1, 2026): > Set `OLLAMA_DEBUG=2` and post the log from start to `inference compute`. ``` time=2026-01-01T09:40:01.222+08:00 level=INFO source=routes.go:1554 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:false 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:D:\\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_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:false ROCR_VISIBLE_DEVICES:]" time=2026-01-01T09:40:01.239+08:00 level=INFO source=images.go:493 msg="total blobs: 38" time=2026-01-01T09:40:01.240+08:00 level=INFO source=images.go:500 msg="total unused blobs removed: 0" time=2026-01-01T09:40:01.241+08:00 level=INFO source=routes.go:1607 msg="Listening on 127.0.0.1:11434 (version 0.13.5)" time=2026-01-01T09:40:01.288+08:00 level=DEBUG source=sched.go:120 msg="starting llm scheduler" time=2026-01-01T09:40:01.289+08:00 level=INFO source=runner.go:67 msg="discovering available GPUs..." time=2026-01-01T09:40:01.297+08:00 level=TRACE source=runner.go:440 msg="starting runner for device discovery" libDirs=[D:\ollama\lib\ollama] extraEnvs=map[] time=2026-01-01T09:40:01.304+08:00 level=INFO source=server.go:429 msg="starting runner" cmd="D:\\ollama\\ollama.exe runner --ollama-engine --port 51152" time=2026-01-01T09:40:01.304+08:00 level=DEBUG source=server.go:430 msg=subprocess CUDA_PATH="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8" CUDA_PATH_V11_6="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8" OLLAMA_DEBUG=2 OLLAMA_MODELS=D:\ollama\models OLLAMA_USE_GPU=1 PATH="D:\\ollama\\lib\\ollama;D:\\Sankyu GitHub\\sankyu chatbot\\Sankyu_Academy_BE\\venv\\Scripts;c:\\Users\\pierr\\AppData\\Roaming\\Code\\User\\globalStorage\\github.copilot-chat\\debugCommand;c:\\Users\\pierr\\AppData\\Roaming\\Code\\User\\globalStorage\\github.copilot-chat\\copilotCli;C:\\Program Files\\Basler\\pylon 7\\Runtime\\x64\\;C:\\Program Files\\Basler\\pylon 7\\Runtime\\Win32\\;C:\\Program Files (x86)\\Common Files\\MVS\\Runtime\\Win32_i86;C:\\Program Files (x86)\\Common Files\\MVS\\Runtime\\Win64_x64;C:\\Python312\\Scripts\\;C:\\Python312\\;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\Scripts\\;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\bin;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\libnvvp;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\extras\\CUPTI\\lib64;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.8\\extras\\CUPTI\\include;C:\\WINDOWS\\system32\\config\\systemprofile\\AppData\\Local\\Microsoft\\WindowsApps;C:\\Users\\pierr\\AppData\\Roaming\\npm;C:\\WINDOWS\\system32\\config\\systemprofile\\.dotnet\\tools;C:\\Users\\pierr\\AppData\\Local\\Programs\\Microsoft VS Code\\bin;C:\\Users\\pierr\\AppData\\Local\\GitHubDesktop\\bin;C:\\Program Files (x86)\\poppler-24.08.0\\Library\\bin;C:\\Program Files\\Git\\bin;C:\\Program Files\\dotnet\\;C:\\poppler-24.08.0\\Library\\bin;C:\\Program Files\\nodejs\\;C:\\ProgramData\\chocolatey\\bin;C:\\Windows\\System32\\nvidia-smi.exe;C:\\Program Files\\NVIDIA Corporation\\NVSMI;C:\\WINDOWS\\system32;C:\\WINDOWS;C:\\WINDOWS\\System32\\Wbem;C:\\WINDOWS\\System32\\WindowsPowerShell\\v1.0\\;C:\\WINDOWS\\System32\\OpenSSH\\;C:\\Program Files\\NVIDIA Corporation\\NVIDIA app\\NvDLISR;;C:\\Program Files\\Docker\\Docker\\resources\\bin;C:\\Users\\pierr\\.local\\bin;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\Scripts\\;C:\\Users\\pierr\\AppData\\Local\\Programs\\Python\\Python310\\;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\bin;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\libnvvp;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\extras\\CUPTI\\lib64;C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6\\extras\\CUPTI\\include;C:\\Users\\pierr\\AppData\\Local\\Microsoft\\WindowsApps;C:\\Users\\pierr\\.dotnet\\tools;C:\\Users\\pierr\\AppData\\Local\\Programs\\Microsoft VS Code\\bin;C:\\Users\\pierr\\AppData\\Local\\GitHubDesktop\\bin;C:\\Users\\pierr\\AppData\\Roaming\\npm;C:\\Users\\pierr\\AppData\\Local\\Programs\\cursor\\resources\\app\\bin;C:\\Users\\pierr\\AppData\\Local\\Programs\\Ollama;D:\\ollama;C:\\Users\\pierr\\Downloads\\ffmpeg-2025-08-11-git-3542260376-full_build\\ffmpeg-2025-08-11-git-3542260376-full_build\\bin;C:\\Users\\pierr\\Downloads\\Release-24.08.0-0 (2)\\Release-24.08.0-0\\poppler-24.08.0\\Library\\bin;c:\\Users\\pierr\\.vscode\\extensions\\ms-python.debugpy-2025.18.0-win32-x64\\bundled\\scripts\\noConfigScripts" OLLAMA_LIBRARY_PATH=D:\ollama\lib\ollama time=2026-01-01T09:40:01.340+08:00 level=INFO source=runner.go:1405 msg="starting ollama engine" time=2026-01-01T09:40:01.353+08:00 level=INFO source=runner.go:1440 msg="Server listening on 127.0.0.1:51152" time=2026-01-01T09:40:01.358+08:00 level=DEBUG source=gguf.go:589 msg=general.architecture type=string time=2026-01-01T09:40:01.359+08:00 level=DEBUG source=gguf.go:589 msg=tokenizer.ggml.model type=string time=2026-01-01T09:40:01.359+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.alignment default=32 time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.alignment default=32 time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.file_type default=0 time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.name default="" time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=general.description default="" time=2026-01-01T09:40:01.360+08:00 level=INFO source=ggml.go:136 msg="" architecture=llama file_type=unknown name="" description="" num_tensors=0 num_key_values=3 time=2026-01-01T09:40:01.360+08:00 level=DEBUG source=ggml.go:94 msg="ggml backend load all from path" path=D:\ollama\lib\ollama load_backend: loaded CPU backend from D:\ollama\lib\ollama\ggml-cpu-alderlake.dll time=2026-01-01T09:40:01.374+08:00 level=INFO source=ggml.go:104 msg=system CPU.0.SSE3=1 CPU.0.SSSE3=1 CPU.0.AVX=1 CPU.0.AVX_VNNI=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(clang) time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.block_count default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.pooling_type default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.expert_count default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.tokens default="&{size:0 values:[]}" time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.scores default="&{size:0 values:[]}" time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.token_type default="&{size:0 values:[]}" time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.merges default="&{size:0 values:[]}" time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.add_bos_token default=true time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.bos_token_id default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.add_eos_token default=false time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.eos_token_id default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.eos_token_ids default="&{size:0 values:[]}" time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=tokenizer.ggml.pre default="" time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.block_count default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.embedding_length default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.head_count default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.head_count_kv default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.key_length default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.rope.dimension_count default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.attention.layer_norm_rms_epsilon default=0 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.rope.freq_base default=100000 time=2026-01-01T09:40:01.374+08:00 level=DEBUG source=ggml.go:282 msg="key with type not found" key=llama.rope.scaling.factor default=1 time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:1380 msg="dummy model load took" duration=16.6606ms time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:1385 msg="gathering device infos took" duration=0s time=2026-01-01T09:40:01.375+08:00 level=TRACE source=runner.go:467 msg="runner enumerated devices" OLLAMA_LIBRARY_PATH=[D:\ollama\lib\ollama] devices=[] time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:437 msg="bootstrap discovery took" duration=78.2778ms OLLAMA_LIBRARY_PATH=[D:\ollama\lib\ollama] extra_envs=map[] time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:124 msg="evaluating which, if any, devices to filter out" initial_count=0 time=2026-01-01T09:40:01.375+08:00 level=TRACE source=runner.go:174 msg="supported GPU library combinations before filtering" supported=map[] time=2026-01-01T09:40:01.375+08:00 level=DEBUG source=runner.go:40 msg="GPU bootstrap discovery took" duration=87.247ms time=2026-01-01T09:40:01.376+08:00 level=INFO source=types.go:60 msg="inference compute" ```
GiteaMirror commented 2026-04-29 09:16:02 -05:00
Author
Owner
Copy Link

@NikolasZaf1 commented on GitHub (Jan 26, 2026):

I had the exact same issue where Ollama was running only on CPU, even though:

  • CUDA was installed correctly
  • NVIDIA GPU was available and visible via nvidia-smi

What I Observed

  • Inside the Ollama lib folder, I could see only one DLL file, instead of multiple CUDA-related DLLs.
  • ollama ps showed the model running on CPU.
  • LLM requests were extremely slow or failing.

Root Cause

The Ollama installation was incomplete or corrupted, so the CUDA backend (ggml-cuda*.dll) was not properly installed or loaded.

Solution (Step-by-Step)

  1. Uninstall Ollama

    • ⚠️ IMPORTANT: When uninstalling, DO NOT check
      “Delete your local models”
      This ensures all downloaded models are preserved.

  2. Reinstall Ollama (latest version)

  3. After reinstall:

    • Verify that the lib folder now contains multiple DLL files, including CUDA-related ones.

  4. Run the following commands:

    ollama list
ollama run <your-model>
ollama ps

After from the last command, you can see that you have usage on GPU. If your problem was resolved, I would appreciate it if you closed the issue.

<!-- gh-comment-id:3801460866 --> @NikolasZaf1 commented on GitHub (Jan 26, 2026): I had **the exact same issue** where Ollama was running **only on CPU**, even though: - CUDA was installed correctly - NVIDIA GPU was available and visible via `nvidia-smi` ### What I Observed - Inside the Ollama `lib` folder, I could see **only one DLL file**, instead of multiple CUDA-related DLLs. - `ollama ps` showed the model running on **CPU**. - LLM requests were extremely slow or failing. ### Root Cause The Ollama installation was **incomplete or corrupted**, so the CUDA backend (`ggml-cuda*.dll`) was not properly installed or loaded. ### Solution (Step-by-Step) 1. **Uninstall Ollama** - ⚠️ **IMPORTANT:** When uninstalling, **DO NOT check** **“Delete your local models”** This ensures all downloaded models are preserved. 2. **Reinstall Ollama** (latest version) 3. After reinstall: - Verify that the `lib` folder now contains **multiple DLL files**, including CUDA-related ones. 4. Run the following commands: ```bash ollama list ollama run <your-model> ollama ps After from the last command, you can see that you have usage on GPU. If your problem was resolved, I would appreciate it if you closed the 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
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#55463
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?