github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 08:59:13 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9765] Confirming if Ollama supports NVIDIA MPS #68440

New Issue
Open
opened 2026-05-04 13:57:40 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-05-04 13:57:40 -05:00
Owner
Copy Link

Originally created by @Dream-Lantern on GitHub (Mar 14, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9765

Hello,
I have enabled MPS mode on my NVIDIA A30 GPU and attempted to start the Ollama service. When I checked using the nvidia-smi command, the type was not showing as "M+C". I know that MPS is working because it functions correctly when I start a training task. I'm not sure if this is due to an error in my operation or if Ollama currently does not support MPS mode.

I would appreciate any guidance on whether I made an error in my operations or if there are known limitations with Ollama and MPS mode.
Thank you very much for your time and assistance. I hope you have a wonderful day!
Best regards,

Here are my operation records and logs:
1.Enable MPS on NVIDIA A30 GPU:
export CUDA_VISIBLE_DEVICES=0 export CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps/ mkdir -p $CUDA_MPS_PIPE_DIRECTORY sudo nvidia-smi -i 0 -c EXCLUSIVE_PROCESS sudo nvidia-cuda-mps-control -d

2.Start Ollama Service and Run model:
sudo env "CUDA_VISIBLE_DEVICES=0" "CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps" ollama serve
sudo env "CUDA_VISIBLE_DEVICES=0" "CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps" ollama run deepseek-r1-qwen:1.5b

3.Check GPU Status with nvidia-smi:
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.183.06 Driver Version: 535.183.06 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 A30 On | 00000000:E1:00.0 Off | Off |
| N/A 78C P0 153W / 165W | 16701MiB / 24576MiB | 100% E. Process |
| | | Disabled |
+-----------------------------------------+----------------------+--------------------------+

+------------------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================|
| 0 N/A N/A 719636 C nvidia-cuda-mps-server 30MiB |
| 0 N/A N/A 726930 M+C python3.8 14622MiB |
| 0 N/A N/A 728422 C ...rs/cuda_v12_avx/ollama_llama_server 2040MiB |
+---------------------------------------------------------------------------------------------+

Logs:
1.mps_control.log
[2025-03-14 18:20:20.981 Control 718283] Accepting connection...
[2025-03-14 18:20:20.981 Control 718283] User did not send valid credentials
[2025-03-14 18:20:20.981 Control 718283] Accepting connection...
[2025-03-14 18:20:20.981 Control 718283] NEW CLIENT 735273 from user 0: Server already exists
[2025-03-14 18:20:21.315 Control 718283] Accepting connection...
[2025-03-14 18:20:21.316 Control 718283] NEW CLIENT 735273 from user 0: Server already exists

2.mps_server.log
[2025-03-14 18:20:20.981 Server 719636] Received new client request
[2025-03-14 18:20:20.981 Server 719636] Worker created
[2025-03-14 18:20:20.981 Server 719636] Creating worker thread
[2025-03-14 18:20:21.316 Server 719636] Received new client request
[2025-03-14 18:20:21.316 Server 719636] Worker created
[2025-03-14 18:20:21.316 Server 719636] Creating worker thread
[2025-03-14 18:20:21.316 Server 719636] Device NVIDIA A30 (uuid 0x47209a07-0x6074dcac-0x7bcb66f2-0x15661a98) is associated
[2025-03-14 18:20:21.316 Server 719636] Status of client {735273, 1} is ACTIVE
[2025-03-14 18:20:21.575 Server 719636] Receive command failed, assuming client exit
[2025-03-14 18:20:21.575 Server 719636] Client {735273, 1} exit
[2025-03-14 18:20:21.575 Server 719636] Client disconnected. Number of active client contexts is 1.

3.ollama_serve.log
2025/03/14 18:06:19 routes.go:1187: INFO server config env="map[CUDA_VISIBLE_DEVICES:0 GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/root/.ollama/models OLLAMA_MULTIUSER_CACHE:false 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://* vscode-webview://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"

Originally created by @Dream-Lantern on GitHub (Mar 14, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9765 Hello, I have enabled MPS mode on my NVIDIA A30 GPU and attempted to start the Ollama service. When I checked using the nvidia-smi command, the type was not showing as "M+C". I know that MPS is working because it functions correctly when I start a training task. I'm not sure if this is due to an error in my operation or if Ollama currently does not support MPS mode. I would appreciate any guidance on whether I made an error in my operations or if there are known limitations with Ollama and MPS mode. Thank you very much for your time and assistance. I hope you have a wonderful day! Best regards, Here are my operation records and logs: 1.Enable MPS on NVIDIA A30 GPU: `export CUDA_VISIBLE_DEVICES=0 export CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps/ mkdir -p $CUDA_MPS_PIPE_DIRECTORY sudo nvidia-smi -i 0 -c EXCLUSIVE_PROCESS sudo nvidia-cuda-mps-control -d` 2.Start Ollama Service and Run model: `sudo env "CUDA_VISIBLE_DEVICES=0" "CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps" ollama serve` `sudo env "CUDA_VISIBLE_DEVICES=0" "CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps" ollama run deepseek-r1-qwen:1.5b` 3.Check GPU Status with nvidia-smi: +---------------------------------------------------------------------------------------+ | NVIDIA-SMI 535.183.06 Driver Version: 535.183.06 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 A30 On | 00000000:E1:00.0 Off | Off | | N/A 78C P0 153W / 165W | 16701MiB / 24576MiB | 100% E. Process | | | | Disabled | +-----------------------------------------+----------------------+--------------------------+ +------------------------------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=======================================================| | 0 N/A N/A 719636 C nvidia-cuda-mps-server 30MiB | | 0 N/A N/A 726930 M+C python3.8 14622MiB | | 0 N/A N/A 728422 C ...rs/cuda_v12_avx/ollama_llama_server 2040MiB | +---------------------------------------------------------------------------------------------+ Logs: 1.mps_control.log [2025-03-14 18:20:20.981 Control 718283] Accepting connection... [2025-03-14 18:20:20.981 Control 718283] User did not send valid credentials [2025-03-14 18:20:20.981 Control 718283] Accepting connection... [2025-03-14 18:20:20.981 Control 718283] NEW CLIENT 735273 from user 0: Server already exists [2025-03-14 18:20:21.315 Control 718283] Accepting connection... [2025-03-14 18:20:21.316 Control 718283] NEW CLIENT 735273 from user 0: Server already exists 2.mps_server.log [2025-03-14 18:20:20.981 Server 719636] Received new client request [2025-03-14 18:20:20.981 Server 719636] Worker created [2025-03-14 18:20:20.981 Server 719636] Creating worker thread [2025-03-14 18:20:21.316 Server 719636] Received new client request [2025-03-14 18:20:21.316 Server 719636] Worker created [2025-03-14 18:20:21.316 Server 719636] Creating worker thread [2025-03-14 18:20:21.316 Server 719636] Device NVIDIA A30 (uuid 0x47209a07-0x6074dcac-0x7bcb66f2-0x15661a98) is associated [2025-03-14 18:20:21.316 Server 719636] Status of client {735273, 1} is ACTIVE [2025-03-14 18:20:21.575 Server 719636] Receive command failed, assuming client exit [2025-03-14 18:20:21.575 Server 719636] Client {735273, 1} exit [2025-03-14 18:20:21.575 Server 719636] Client disconnected. Number of active client contexts is 1. 3.ollama_serve.log 2025/03/14 18:06:19 routes.go:1187: INFO server config env="map[CUDA_VISIBLE_DEVICES:0 GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://127.0.0.1:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/root/.ollama/models OLLAMA_MULTIUSER_CACHE:false 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://* vscode-webview://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"
GiteaMirror added the feature request label 2026-05-04 13:57:40 -05:00
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 feature request
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#68440
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?