github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-09 01:59:50 -05:00
Code Issues 14.6k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #8546] I have multiple GPUs, but I cannot use them all。 #5514

New Issue
Closed
opened 2026-04-12 16:45:57 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-12 16:45:57 -05:00
Owner
Copy Link

Originally created by @dayphosphor on GitHub (Jan 23, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8546

What is the issue?

I have four 4090 GPUs.

Image

but, only one GPU is used.

I have configured the environment variables。 /etc/systemd/system/ollama.service

Image

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @dayphosphor on GitHub (Jan 23, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8546 ### What is the issue? I have four 4090 GPUs. <img width="815" alt="Image" src="https://github.com/user-attachments/assets/6edf430c-a927-4010-a46b-6ffaa8288f30" /> but, only one GPU is used. I have configured the environment variables。 /etc/systemd/system/ollama.service <img width="823" alt="Image" src="https://github.com/user-attachments/assets/2c923009-c235-49cc-a126-25527ca964c2" /> ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the bug label 2026-04-12 16:45:57 -05:00
GiteaMirror commented 2026-04-12 16:45:59 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jan 23, 2025):

If the model fits on one GPU, that's all that will be used - multiple GPUs doesn't confer a performance increase, see here. If you want to force ollama to use all GPUs for a model, set OLLAMA_SCHED_SPREAD.

<!-- gh-comment-id:2609069452 --> @rick-github commented on GitHub (Jan 23, 2025): If the model fits on one GPU, that's all that will be used - multiple GPUs doesn't confer a performance increase, see [here](https://github.com/ollama/ollama/issues/7648#issuecomment-2473561990). If you want to force ollama to use all GPUs for a model, set [`OLLAMA_SCHED_SPREAD`](https://github.com/ollama/ollama/blob/ca2f9843c8c71491d5abf626c73508e5a1685cea/envconfig/config.go#L251).
GiteaMirror commented 2026-04-12 16:46:00 -05:00
Author
Owner
Copy Link

@UberMetroid commented on GitHub (Jan 23, 2025):

This is solved and should be marked closed.

<!-- gh-comment-id:2609828387 --> @UberMetroid commented on GitHub (Jan 23, 2025): This is solved and should be marked closed.
GiteaMirror commented 2026-04-12 16:46:01 -05:00
Author
Owner
Copy Link

@yxc0915 commented on GitHub (Jan 23, 2025):

40系又不支持SLI，如果是3090ti兴许还能多卡调度

<!-- gh-comment-id:2610405561 --> @yxc0915 commented on GitHub (Jan 23, 2025): 40系又不支持SLI，如果是3090ti兴许还能多卡调度
GiteaMirror commented 2026-04-12 16:46:02 -05:00
Author
Owner
Copy Link

@UberMetroid commented on GitHub (Jan 24, 2025):

40系又不支持SLI，如果是3090ti兴许还能多卡调度

With OLLAMA_SCHED_SPREAD you dont need SLI for multi use GPUs.

<!-- gh-comment-id:2612482930 --> @UberMetroid commented on GitHub (Jan 24, 2025): > 40系又不支持SLI，如果是3090ti兴许还能多卡调度 With [OLLAMA_SCHED_SPREAD](https://github.com/ollama/ollama/blob/ca2f9843c8c71491d5abf626c73508e5a1685cea/envconfig/config.go#L251) you dont need SLI for multi use GPUs.
GiteaMirror commented 2026-04-12 16:46:03 -05:00
Author
Owner
Copy Link

@mmdyu commented on GitHub (Feb 15, 2025):

If the model fits on one GPU, that's all that will be used - multiple GPUs doesn't confer a performance increase, see here. If you want to force ollama to use all GPUs for a model, set OLLAMA_SCHED_SPREAD.

After adding, the total number of GPUs should not exceed 100%

Image

<!-- gh-comment-id:2661005785 --> @mmdyu commented on GitHub (Feb 15, 2025): > If the model fits on one GPU, that's all that will be used - multiple GPUs doesn't confer a performance increase, see [here](https://github.com/ollama/ollama/issues/7648#issuecomment-2473561990). If you want to force ollama to use all GPUs for a model, set [`OLLAMA_SCHED_SPREAD`](https://github.com/ollama/ollama/blob/ca2f9843c8c71491d5abf626c73508e5a1685cea/envconfig/config.go#L251). After adding, the total number of GPUs should not exceed 100% ![Image](https://github.com/user-attachments/assets/d4d84e13-9591-44bc-8c7a-056ed6951f59)
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#5514
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?