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 #3536] Why my ollama not use gpu？ #64218

New Issue
Closed
opened 2026-05-03 16:38:30 -05:00 by GiteaMirror · 9 comments
GiteaMirror commented 2026-05-03 16:38:30 -05:00
Owner
Copy Link

Originally created by @loks666 on GitHub (Apr 8, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/3536

My ollama is use windows installer setup running.There was a problem,when I watch my tsak manager,I noticed that my gpu was not being used. I want know that's why? or say I need run what command?
I wish someone could solve my problem，thanks
image

Originally created by @loks666 on GitHub (Apr 8, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/3536 My ollama is use windows installer setup running.There was a problem,when I watch my tsak manager,I noticed that my gpu was not being used. I want know that's why? or say I need run what command? I wish someone could solve my problem，thanks ![image](https://github.com/ollama/ollama/assets/33214035/1712abc8-33ba-441e-b6e4-ec7d6a34d662)
GiteaMirror added the question label 2026-05-03 16:38:30 -05:00
GiteaMirror commented 2026-05-03 16:38:34 -05:00
Author
Owner
Copy Link

@aosan commented on GitHub (Apr 8, 2024):

What model are you using? I can see your memory is at 95%. GPU is fully utilised by models fitting in VRAM, models using under 11 GB would fit in your 2080Ti VRAM.

GPU usage would show up when you make a request, e.g. ollama run mistral and make a request: "why is the sky blue?"

GPU load would appear while the model is providing the response.

<!-- gh-comment-id:2042243856 --> @aosan commented on GitHub (Apr 8, 2024): What model are you using? I can see your memory is at 95%. GPU is fully utilised by models fitting in VRAM, models using under 11 GB would fit in your 2080Ti VRAM. GPU usage would show up when you make a request, e.g. `ollama run mistral` and make a request: "why is the sky blue?" GPU load would appear while the model is providing the response.
GiteaMirror commented 2026-05-03 16:38:35 -05:00
Author
Owner
Copy Link

@loks666 commented on GitHub (Apr 8, 2024):

Are you saying that my computer will only use the gpu if I run a model that is smaller than my gpu memory?

<!-- gh-comment-id:2042248843 --> @loks666 commented on GitHub (Apr 8, 2024): Are you saying that my computer will only use the gpu if I run a model that is smaller than my gpu memory?
GiteaMirror commented 2026-05-03 16:38:36 -05:00
Author
Owner
Copy Link

@aosan commented on GitHub (Apr 8, 2024):

Depends on the model, some models offload layers to GPU even if they can't fit in VRAM and are running on CPU.

For maximum performance, any model fitting in VRAM will benefit from GPU processing.

<!-- gh-comment-id:2042407607 --> @aosan commented on GitHub (Apr 8, 2024): Depends on the model, some models offload layers to GPU even if they can't fit in VRAM and are running on CPU. For maximum performance, any model fitting in VRAM will benefit from GPU processing.
GiteaMirror commented 2026-05-03 16:38:39 -05:00
Author
Owner
Copy Link

@loks666 commented on GitHub (Apr 11, 2024):

So when I install ollama using Windows operating system programs, the default is to use the gpu to run large models? I don't need to do anything extra?

<!-- gh-comment-id:2050028784 --> @loks666 commented on GitHub (Apr 11, 2024): So when I install ollama using Windows operating system programs, the default is to use the gpu to run large models? I don't need to do anything extra?
GiteaMirror commented 2026-05-03 16:38:41 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Apr 12, 2024):

Hi @loks666, yes, it will try to offload as many "layers" of the model onto your GPU as possible. If the model is bigger than the memory of your GPU, those extra layers will be run on the CPU instead. You can try running smaller models, or use a more quantized version of the same model.

I'm going to go ahead and close the issue, but feel free to keep commenting.

<!-- gh-comment-id:2052531436 --> @pdevine commented on GitHub (Apr 12, 2024): Hi @loks666, yes, it will try to offload as many "layers" of the model onto your GPU as possible. If the model is bigger than the memory of your GPU, those extra layers will be run on the CPU instead. You can try running smaller models, or use a more quantized version of the same model. I'm going to go ahead and close the issue, but feel free to keep commenting.
GiteaMirror commented 2026-05-03 16:38:44 -05:00
Author
Owner
Copy Link

@highbrow-228 commented on GitHub (Mar 17, 2025):

It doesn't help, at least in my case. I have a video card with a 20GB GPU. Cuda, nvidia toolkit are installed. However, the models do not use the GPU at all😩. I have tried llama3.2 as the small model that will fit and the big one (llama3.3).

<!-- gh-comment-id:2728760284 --> @highbrow-228 commented on GitHub (Mar 17, 2025): It doesn't help, at least in my case. I have a video card with a 20GB GPU. Cuda, nvidia toolkit are installed. However, the models do not use the GPU at all😩. I have tried llama3.2 as the small model that will fit and the big one (llama3.3).
GiteaMirror commented 2026-05-03 16:38:46 -05:00
Author
Owner
Copy Link

@Mohamed0Hegazi commented on GitHub (Mar 17, 2025):

بايثون اين للكتاب

<!-- gh-comment-id:2728783646 --> @Mohamed0Hegazi commented on GitHub (Mar 17, 2025): بايثون اين للكتاب
GiteaMirror commented 2026-05-03 16:38:48 -05:00
Author
Owner
Copy Link

@Mohamed0Hegazi commented on GitHub (Mar 17, 2025):

هل انتهيت من الكتاب
#بايثون
#الما

<!-- gh-comment-id:2728788074 --> @Mohamed0Hegazi commented on GitHub (Mar 17, 2025): هل انتهيت من الكتاب #بايثون #الما
GiteaMirror commented 2026-05-03 16:38:49 -05:00
Author
Owner
Copy Link

@Mohamed0Hegazi commented on GitHub (Mar 17, 2025):

لقد طلبت من بايثون صناعه الكتاب الالكتروني تنميه الذات

<!-- gh-comment-id:2728835977 --> @Mohamed0Hegazi commented on GitHub (Mar 17, 2025): لقد طلبت من بايثون صناعه الكتاب الالكتروني تنميه الذات
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 question
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#64218
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?