github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #9002] How to limit GPU memory usage for Ollama when running large models? #67905

New Issue
Closed
opened 2026-05-04 12:00:04 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-05-04 12:00:04 -05:00
Owner
Copy Link

Originally created by @lizheyong on GitHub (Feb 11, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9002

Environment

  • OS: CentOS (offline environment)
  • Hardware: 2x NVIDIA A100 40GB GPUs
  • Model: Deepseek-r1-70B (43GB)
  • Current Configuration:
    • CUDA_VISIBLE_DEVICES=0,1
    • OLLAMA_MAX_LOADED_MODELS=1

Current Behavior

  • The model spans across both GPUs
  • Total GPU memory usage is around 80GB
  • Both GPUs are fully occupied

Expected Behavior

  • The model should use partial memory of the second GPU
  • Reserve some GPU memory on second GPU for other deep learning tasks
  • Still maintain model functionality

What I've Tried

  • Checked Ollama documentation
  • Searched online for solutions
  • Modified ollama.service configuration

Question

Is there any way to effectively limit Ollama's GPU memory usage, particularly on the second GPU?

Originally created by @lizheyong on GitHub (Feb 11, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9002 ## Environment - OS: CentOS (offline environment) - Hardware: 2x NVIDIA A100 40GB GPUs - Model: Deepseek-r1-70B (43GB) - Current Configuration: - `CUDA_VISIBLE_DEVICES=0,1` - `OLLAMA_MAX_LOADED_MODELS=1` ## Current Behavior - The model spans across both GPUs - Total GPU memory usage is around 80GB - Both GPUs are fully occupied ## Expected Behavior - The model should use partial memory of the second GPU - Reserve some GPU memory on second GPU for other deep learning tasks - Still maintain model functionality ## What I've Tried - Checked Ollama documentation - Searched online for solutions - Modified ollama.service configuration ## Question Is there any way to effectively limit Ollama's GPU memory usage, particularly on the second GPU?
GiteaMirror added the question label 2026-05-04 12:00:05 -05:00
GiteaMirror commented 2026-05-04 12:00:08 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 11, 2025):

https://github.com/ollama/ollama/issues/8597#issuecomment-2614533288

Or reduce num_ctx.

<!-- gh-comment-id:2650215196 --> @rick-github commented on GitHub (Feb 11, 2025): https://github.com/ollama/ollama/issues/8597#issuecomment-2614533288 Or reduce `num_ctx`.
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#67905
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?