github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 16:40:08 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #13765] Randomly uses CPU even when GPU available until I restart #71081

New Issue
Closed
opened 2026-05-04 23:57:16 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-05-04 23:57:16 -05:00
Owner
Copy Link

Originally created by @CooperCorona on GitHub (Jan 17, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/13765

What is the issue?

Occasionally, ollama will refuse to use my GPU. nvidia-smi shows I have plenty of available VRAM. But ollama run will load the model entirely on the CPU, with ollama ps showing 100% CPU usage. This happens seemingly randomly. I'll go days using ollama just fine, then the next time I run it, all CPU.

I'm running 0.13.5 via OpenWebUI's Ollama Docker image, which is at version v0.7.2, on Ubuntu 22.04.5 LTS. I notice this via OpenWebUI, which fails with 500: model requires more system memory (18.9 GiB) than is available (13.1 GiB) (because it's using CPU, not GPU). I have OpenWebUI to refuse to use the CPU, because before, it would silently use the CPU and just generate at 10% speed. This has occurred throughout OpenWebUI and Ollama versions.

Let me know if you need more information or would like me to do any debugging. I solve this just by restarting the Docker container, but it's quite annoying.

Relevant log output

OS

Docker

GPU

Nvidia

CPU

AMD

Ollama version

0.13.5

Originally created by @CooperCorona on GitHub (Jan 17, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/13765 ### What is the issue? Occasionally, ollama will refuse to use my GPU. `nvidia-smi` shows I have plenty of available VRAM. But `ollama run` will load the model entirely on the CPU, with `ollama ps` showing 100% CPU usage. This happens seemingly randomly. I'll go days using ollama just fine, then the next time I run it, all CPU. I'm running 0.13.5 via OpenWebUI's Ollama Docker image, which is at version v0.7.2, on Ubuntu 22.04.5 LTS. I notice this via OpenWebUI, which fails with `500: model requires more system memory (18.9 GiB) than is available (13.1 GiB)` (because it's using CPU, not GPU). I have OpenWebUI to refuse to use the CPU, because before, it would silently use the CPU and just generate at 10% speed. This has occurred throughout OpenWebUI and Ollama versions. Let me know if you need more information or would like me to do any debugging. I solve this just by restarting the Docker container, but it's quite annoying. ### Relevant log output ```shell ``` ### OS Docker ### GPU Nvidia ### CPU AMD ### Ollama version 0.13.5
GiteaMirror added the bugnvidia labels 2026-05-04 23:57:17 -05:00
GiteaMirror commented 2026-05-04 23:57:24 -05:00
Author
Owner
Copy Link

@mchiang0610 commented on GitHub (Jan 18, 2026):

Thank you for reporting this. May I ask if this issue exists on the latest Ollama version? v0.14.2

I'd love to know which GPU you have too.

<!-- gh-comment-id:3764942138 --> @mchiang0610 commented on GitHub (Jan 18, 2026): Thank you for reporting this. May I ask if this issue exists on the latest Ollama version? v0.14.2 I'd love to know which GPU you have too.
GiteaMirror commented 2026-05-04 23:57:26 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 18, 2026):

https://github.com/ollama/ollama/blob/main/docs/troubleshooting.mdx#linux-docker

<!-- gh-comment-id:3765078195 --> @rick-github commented on GitHub (Jan 18, 2026): https://github.com/ollama/ollama/blob/main/docs/troubleshooting.mdx#linux-docker
GiteaMirror commented 2026-05-04 23:57:27 -05:00
Author
Owner
Copy Link

@thomas-meier85 commented on GitHub (Jan 18, 2026):

Try adding

{
    "runtimes": {
        "nvidia": {
            "args": [],
            "path": "nvidia-container-runtime"
        }
    },
        "exec-opts": ["native.cgroupdriver=cgroupfs"]
}

to etc/docker/daemon.json

Restart your machine and check if the error still happens.

<!-- gh-comment-id:3765151503 --> @thomas-meier85 commented on GitHub (Jan 18, 2026): Try adding ``` { "runtimes": { "nvidia": { "args": [], "path": "nvidia-container-runtime" } }, "exec-opts": ["native.cgroupdriver=cgroupfs"] } ``` to etc/docker/daemon.json Restart your machine and check if the error still happens.
GiteaMirror commented 2026-05-04 23:57:27 -05:00
Author
Owner
Copy Link

@CooperCorona commented on GitHub (Mar 15, 2026):

sorry, completely missed all the responses here. This just happened again on Ollama v0.17.2. My GPU is an EVGA RTX 3090 and Nvidia FE 3090.

I'll try thomas' suggestion and see if it happens again. Since this bug is stale, I'll only reopen it if it occurs yet again.

<!-- gh-comment-id:4063564479 --> @CooperCorona commented on GitHub (Mar 15, 2026): sorry, completely missed all the responses here. This just happened again on Ollama v0.17.2. My GPU is an EVGA RTX 3090 and Nvidia FE 3090. I'll try thomas' suggestion and see if it happens again. Since this bug is stale, I'll only reopen it if it occurs yet again.
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 nvidia
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#71081
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?