github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 05:05:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #6529] Ollama will stop using GPU when the total graphics memory usage exceeds the dedicated graphics memory size #4111

New Issue
Closed
opened 2026-04-12 15:00:37 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 15:00:37 -05:00
Owner
Copy Link

Originally created by @Eomfped on GitHub (Aug 27, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6529

What is the issue?

When I tried to run models, I noticed that once the total graphics memory usage exceeded the dedicated graphics memory size of the graphics card, Ollama would stop using GPU for inference.
I saw in the task manager that the utilization rate of the graphics card was 0%. But when I tried to use a low graphics memory usage model, the GPU utilization increased to 100%.

OS

Windows

GPU

Nvidia

CPU

Intel

Ollama version

0.3.6

Originally created by @Eomfped on GitHub (Aug 27, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6529 ### What is the issue? When I tried to run models, I noticed that once the total graphics memory usage exceeded the dedicated graphics memory size of the graphics card, Ollama would stop using GPU for inference. I saw in the task manager that the utilization rate of the graphics card was 0%. But when I tried to use a low graphics memory usage model, the GPU utilization increased to 100%. ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version 0.3.6
GiteaMirror added the bug label 2026-04-12 15:00:37 -05:00
GiteaMirror commented 2026-04-12 15:00:39 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Aug 27, 2024):

Server logs will give more insight into what is happening. but generally, ollama will split the model between the GPU and CPU, loading as much as it can on the GPU. However, the CPU is less efficient than the GPU, so inference of the layers on the CPU will take longer than the layers on the GPU. This means there will be periods where the GPU is idle, waiting for the CPU inference to complete.

<!-- gh-comment-id:2313341643 --> @rick-github commented on GitHub (Aug 27, 2024): [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) will give more insight into what is happening. but generally, ollama will split the model between the GPU and CPU, loading as much as it can on the GPU. However, the CPU is less efficient than the GPU, so inference of the layers on the CPU will take longer than the layers on the GPU. This means there will be periods where the GPU is idle, waiting for the CPU inference to complete.
GiteaMirror commented 2026-04-12 15:00:39 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Aug 28, 2024):

@UserGzy @rick-github 's answer is correct, and Ollama is behaving as intended. Any time you split a model across GPU/CPU it will slow down because inference needs to wait for the CPU which will cause the GPU to idle.

I'll go ahead and close out the issue.

<!-- gh-comment-id:2316221639 --> @pdevine commented on GitHub (Aug 28, 2024): @UserGzy @rick-github 's answer is correct, and Ollama is behaving as intended. Any time you split a model across GPU/CPU it will slow down because inference needs to wait for the CPU which will cause the GPU to idle. I'll go ahead and close out 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#4111
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?