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 #9310] 8 GPUs want to start 8 same models #6075

New Issue
Closed
opened 2026-04-12 17:24:14 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 17:24:14 -05:00
Owner
Copy Link

Originally created by @AltenLi on GitHub (Feb 24, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9310

I tried so many methods.
environment: win10 22H, latest, nightly

methods:
1.multi ollama serve: failed on gpu split.
2.multi ollama docker: --gpus all & CUDA_VISIBLE_DEVICES=0 / --gpus all & CUDA_VISIBLE_DEVICES=1 and so on.. use local different copied model path into docker (to save download time), load model extremely slow, 32B used 15min. and often killed by gpu vram..

need:
multi same model server locally.

Thanks!!!

Originally created by @AltenLi on GitHub (Feb 24, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9310 I tried so many methods. environment: win10 22H, latest, nightly methods: 1.multi ollama serve: failed on gpu split. 2.multi ollama docker: --gpus all & CUDA_VISIBLE_DEVICES=0 / --gpus all & CUDA_VISIBLE_DEVICES=1 and so on.. use local different copied model path into docker (to save download time), load model extremely slow, 32B used 15min. and often killed by gpu vram.. need: multi same model server locally. Thanks!!!
GiteaMirror added the feature request label 2026-04-12 17:24:14 -05:00
GiteaMirror commented 2026-04-12 17:24:16 -05:00
Author
Owner
Copy Link

@ShadovvSinger commented on GitHub (Feb 25, 2025):

see https://github.com/ollama/ollama/issues/7206#issuecomment-2413928300
you can try: start multiple ollama on different GPU and different port, use nginx to to distribute the requests

I am also very thirsty to this feature (one ollama, multiple copies on multiple GPUs)
please follow https://github.com/ollama/ollama/issues/3902

<!-- gh-comment-id:2682218932 --> @ShadovvSinger commented on GitHub (Feb 25, 2025): see https://github.com/ollama/ollama/issues/7206#issuecomment-2413928300 you can try: start multiple ollama on different GPU and different port, use nginx to to distribute the requests I am also very thirsty to this feature (one ollama, multiple copies on multiple GPUs) please follow https://github.com/ollama/ollama/issues/3902
GiteaMirror commented 2026-04-12 17:24:17 -05:00
Author
Owner
Copy Link

@ShadovvSinger commented on GitHub (Feb 25, 2025):

"1.multi ollama serve: failed on gpu split."
I have successfully deployed using this method.
I use screen to start different terminal, set CUDA_VISIBLE_DEVICES=0/1/2/3/4... (the CUDA number)
OLLAMA_HOST= different port
I don't know if windows support this method, but it should work some how.

<!-- gh-comment-id:2682241127 --> @ShadovvSinger commented on GitHub (Feb 25, 2025): "1.multi ollama serve: failed on gpu split." I have successfully deployed using this method. I use screen to start different terminal, set CUDA_VISIBLE_DEVICES=0/1/2/3/4... (the CUDA number) OLLAMA_HOST= different port I don't know if windows support this method, but it should work some how.
GiteaMirror commented 2026-04-12 17:24:18 -05:00
Author
Owner
Copy Link

@wenhui-ml commented on GitHub (May 4, 2025):

hi，is there any progress to run one same llm model on multi gpus on windows, in order to serve distributed request?

<!-- gh-comment-id:2849148816 --> @wenhui-ml commented on GitHub (May 4, 2025): hi，is there any progress to run one same llm model on multi gpus on windows, in order to serve distributed request?
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#6075
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?