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 #5493] unable to load nvcuda #3436

New Issue
Closed
opened 2026-04-12 14:05:40 -05:00 by GiteaMirror · 7 comments
GiteaMirror commented 2026-04-12 14:05:40 -05:00
Owner
Copy Link

Originally created by @yake-cyber on GitHub (Jul 5, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5493

What is the issue?

my ollama does not run on the NVIDIA gpu and i use the debug mode and find this message
"time=2024-07-04T17:13:20.134+08:00 level=DEBUG source=gpu.go:385 msg="Unable to load nvcuda" library=/usr/lib/libcuda.so.418.74 error="Unable to load /usr/lib/libcuda.so.418.74 library to query for Nvidia GPUs: /usr/lib/libcuda.so.418.74: wrong ELF class: ELFCLASS32"
dlerr: /usr/lib64/libcuda.so.418.74: undefined symbol: cuCtxCreate_v3
time=2024-07-04T17:13:20.135+08:00 level=DEBUG source=gpu.go:385 msg="Unable to load nvcuda" library=/usr/lib64/libcuda.so.418.74 error="symbol lookup for cuCtxCreate_v3 failed: /usr/lib64/libcuda.so.418.74: undefined symbol: cuCtxCreate_v3"
time=2024-07-04T17:13:20.135+08:00 level=DEBUG source=gpu.go:304 msg="Searching for GPU library" name=libcudart.so*"
could you please clarify what this means and how can i resolving this issue?
debug-ollama.txt
nvidia-smi
log20240704.txt

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.1.40

Originally created by @yake-cyber on GitHub (Jul 5, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5493 ### What is the issue? my ollama does not run on the NVIDIA gpu and i use the debug mode and find this message "time=2024-07-04T17:13:20.134+08:00 level=DEBUG source=gpu.go:385 msg="Unable to load nvcuda" library=/usr/lib/libcuda.so.418.74 error="Unable to load /usr/lib/libcuda.so.418.74 library to query for Nvidia GPUs: /usr/lib/libcuda.so.418.74: wrong ELF class: ELFCLASS32" dlerr: /usr/lib64/libcuda.so.418.74: undefined symbol: cuCtxCreate_v3 time=2024-07-04T17:13:20.135+08:00 level=DEBUG source=gpu.go:385 msg="Unable to load nvcuda" library=/usr/lib64/libcuda.so.418.74 error="symbol lookup for cuCtxCreate_v3 failed: /usr/lib64/libcuda.so.418.74: undefined symbol: cuCtxCreate_v3" time=2024-07-04T17:13:20.135+08:00 level=DEBUG source=gpu.go:304 msg="Searching for GPU library" name=libcudart.so*" could you please clarify what this means and how can i resolving this issue? [debug-ollama.txt](https://github.com/user-attachments/files/16104245/debug-ollama.txt) ![nvidia-smi](https://github.com/ollama/ollama/assets/174697336/16f66b47-fa5c-4be6-a748-7e5173e69279) [log20240704.txt](https://github.com/user-attachments/files/16104248/log20240704.txt) ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.1.40
GiteaMirror added the bug label 2026-04-12 14:05:40 -05:00
GiteaMirror commented 2026-04-12 14:05:43 -05:00
Author
Owner
Copy Link

@yake-cyber commented on GitHub (Jul 5, 2024):

cat /proc/driver/nvidia/version,and the result is
NVRM version: NVIDIA UNIX x86_64 Kernel Module 418.74 Wed May 1 11:49:41 CDT 2019

<!-- gh-comment-id:2210251365 --> @yake-cyber commented on GitHub (Jul 5, 2024): cat /proc/driver/nvidia/version,and the result is NVRM version: NVIDIA UNIX x86_64 Kernel Module 418.74 Wed May 1 11:49:41 CDT 2019
GiteaMirror commented 2026-04-12 14:05:44 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jul 11, 2024):

Hi there, I believe you'll need Cuda 11 or later for Ollama – installing the latest Nvidia drivers for your graphics card should fix this.

<!-- gh-comment-id:2222347725 --> @jmorganca commented on GitHub (Jul 11, 2024): Hi there, I believe you'll need Cuda 11 or later for Ollama – installing the latest Nvidia drivers for your graphics card should fix this.
GiteaMirror commented 2026-04-12 14:05:46 -05:00
Author
Owner
Copy Link

@yake-cyber commented on GitHub (Jul 11, 2024):

update nvidia driver from 418.74 to 515.105.01
update cuda from 10.1 to 11.7

<!-- gh-comment-id:2222348460 --> @yake-cyber commented on GitHub (Jul 11, 2024): update nvidia driver from 418.74 to 515.105.01 update cuda from 10.1 to 11.7
GiteaMirror commented 2026-04-12 14:05:47 -05:00
Author
Owner
Copy Link

@birkhoff2017 commented on GitHub (Jul 24, 2024):

update nvidia driver from 418.74 to 515.105.01 update cuda from 10.1 to 11.7

does update cuda version solved this problem?

<!-- gh-comment-id:2247462892 --> @birkhoff2017 commented on GitHub (Jul 24, 2024): > update nvidia driver from 418.74 to 515.105.01 update cuda from 10.1 to 11.7 does update cuda version solved this problem?
GiteaMirror commented 2026-04-12 14:05:48 -05:00
Author
Owner
Copy Link

@thomasWos commented on GitHub (Aug 2, 2024):

same error on my laptop:

time=2024-08-02T12:29:29.168+10:00 level=INFO source=routes.go:1156 msg="Listening on 127.0.0.1:11434 (version 0.3.2)"
time=2024-08-02T12:29:29.170+10:00 level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cpu cpu_avx cpu_avx2 cuda_v11.3 rocm_v6.1]"
time=2024-08-02T12:29:29.170+10:00 level=INFO source=gpu.go:205 msg="looking for compatible GPUs"
time=2024-08-02T12:29:29.208+10:00 level=INFO source=gpu.go:564 msg="unable to load cuda driver library" library=C:\WINDOWS\system32\nvcuda.dll error="symbol lookup for cuCtxCreate_v3 failed: The specified procedure could not be found.\r\n"
time=2024-08-02T12:29:29.222+10:00 level=INFO source=gpu.go:346 msg="no compatible GPUs were discovered"

CUDA Version: 11.0

<!-- gh-comment-id:2264396628 --> @thomasWos commented on GitHub (Aug 2, 2024): same error on my laptop: time=2024-08-02T12:29:29.168+10:00 level=INFO source=routes.go:1156 msg="Listening on 127.0.0.1:11434 (version 0.3.2)" time=2024-08-02T12:29:29.170+10:00 level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cpu cpu_avx cpu_avx2 cuda_v11.3 rocm_v6.1]" time=2024-08-02T12:29:29.170+10:00 level=INFO source=gpu.go:205 msg="looking for compatible GPUs" time=2024-08-02T12:29:29.208+10:00 level=INFO source=gpu.go:564 msg="unable to load cuda driver library" library=C:\WINDOWS\system32\nvcuda.dll error="symbol lookup for cuCtxCreate_v3 failed: The specified procedure could not be found.\r\n" time=2024-08-02T12:29:29.222+10:00 level=INFO source=gpu.go:346 msg="no compatible GPUs were discovered" CUDA Version: 11.0
GiteaMirror commented 2026-04-12 14:05:49 -05:00
Author
Owner
Copy Link

@Yuhao-Luo commented on GitHub (Aug 12, 2024):

same error on p40
NVIDIA-SMI 460.73.01
CUDA Version: 11.2

> ollama serve
2024/08/12 19:46:02 routes.go:1033: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_LLM_LIBRARY: OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_VRAM:0 OLLAMA_MODELS:/root/.ollama/models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 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://*] OLLAMA_RUNNERS_DIR: OLLAMA_SCHED_SPREAD:false OLLAMA_TMPDIR: ROCR_VISIBLE_DEVICES:]"
time=2024-08-12T19:46:02.378+08:00 level=INFO source=images.go:751 msg="total blobs: 0"
time=2024-08-12T19:46:02.378+08:00 level=INFO source=images.go:758 msg="total unused blobs removed: 0"
time=2024-08-12T19:46:02.379+08:00 level=INFO source=routes.go:1080 msg="Listening on 127.0.0.1:11434 (version 0.2.1)"
time=2024-08-12T19:46:02.379+08:00 level=INFO source=payload.go:30 msg="extracting embedded files" dir=/tmp/ollama3129743963/runners
time=2024-08-12T19:46:07.169+08:00 level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cpu_avx2 cuda_v11 rocm_v60101 cpu cpu_avx]"
time=2024-08-12T19:46:07.169+08:00 level=INFO source=gpu.go:205 msg="looking for compatible GPUs"
time=2024-08-12T19:46:07.176+08:00 level=INFO source=gpu.go:534 msg="unable to load cuda driver library" library=/usr/lib/x86_64-linux-gnu/libcuda.so.460.73.01 error="symbol lookup for cuCtxCreate_v3 failed: /usr/lib/x86_64-linux-gnu/libcuda.so.460.73.01: undefined symbol: cuCtxCreate_v3"
time=2024-08-12T19:46:07.676+08:00 level=INFO source=types.go:103 msg="inference compute" id=GPU-de51333b-d42b-ed7c-98c6-29465e69005d library=cuda compute=6.1 driver=0.0 name="" total="22.4 GiB" available="22.2 GiB"
<!-- gh-comment-id:2283753492 --> @Yuhao-Luo commented on GitHub (Aug 12, 2024): same error on p40 NVIDIA-SMI 460.73.01 CUDA Version: 11.2 ``` > ollama serve 2024/08/12 19:46:02 routes.go:1033: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_LLM_LIBRARY: OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_VRAM:0 OLLAMA_MODELS:/root/.ollama/models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 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://*] OLLAMA_RUNNERS_DIR: OLLAMA_SCHED_SPREAD:false OLLAMA_TMPDIR: ROCR_VISIBLE_DEVICES:]" time=2024-08-12T19:46:02.378+08:00 level=INFO source=images.go:751 msg="total blobs: 0" time=2024-08-12T19:46:02.378+08:00 level=INFO source=images.go:758 msg="total unused blobs removed: 0" time=2024-08-12T19:46:02.379+08:00 level=INFO source=routes.go:1080 msg="Listening on 127.0.0.1:11434 (version 0.2.1)" time=2024-08-12T19:46:02.379+08:00 level=INFO source=payload.go:30 msg="extracting embedded files" dir=/tmp/ollama3129743963/runners time=2024-08-12T19:46:07.169+08:00 level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cpu_avx2 cuda_v11 rocm_v60101 cpu cpu_avx]" time=2024-08-12T19:46:07.169+08:00 level=INFO source=gpu.go:205 msg="looking for compatible GPUs" time=2024-08-12T19:46:07.176+08:00 level=INFO source=gpu.go:534 msg="unable to load cuda driver library" library=/usr/lib/x86_64-linux-gnu/libcuda.so.460.73.01 error="symbol lookup for cuCtxCreate_v3 failed: /usr/lib/x86_64-linux-gnu/libcuda.so.460.73.01: undefined symbol: cuCtxCreate_v3" time=2024-08-12T19:46:07.676+08:00 level=INFO source=types.go:103 msg="inference compute" id=GPU-de51333b-d42b-ed7c-98c6-29465e69005d library=cuda compute=6.1 driver=0.0 name="" total="22.4 GiB" available="22.2 GiB" ```
GiteaMirror commented 2026-04-12 14:05:50 -05:00
Author
Owner
Copy Link

@thomasWos commented on GitHub (Aug 13, 2024):

I have upgraded my drivers, CUDA Version: 12.2.
Works fine since

<!-- gh-comment-id:2285281010 --> @thomasWos commented on GitHub (Aug 13, 2024): I have upgraded my drivers, CUDA Version: 12.2. Works fine since
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#3436
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?