[GH-ISSUE #2826] How to enable GPU support on Ubuntu? #1717

Closed
opened 2026-04-12 11:41:40 -05:00 by GiteaMirror · 2 comments
Owner

Originally created by @realcarlos on GitHub (Feb 29, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2826

Hi , I am using a DGX H800 server , and when I run

ollama run llama2:70b

I found that model is loaded into memory , not GPU .

Screenshot 2024-02-29 at 12 03 28
Originally created by @realcarlos on GitHub (Feb 29, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2826 Hi , I am using a DGX H800 server , and when I run `ollama run llama2:70b` I found that model is loaded into memory , not GPU . <img width="558" alt="Screenshot 2024-02-29 at 12 03 28" src="https://github.com/ollama/ollama/assets/13853/182fb3d8-f075-4c33-8d42-82cec09b5459">
Author
Owner

@tylinux commented on GitHub (Feb 29, 2024):

Maybe vram is not enough to load model, run OLLAMA_DEBUG=1 ollama serve, than run your model, see if there have not enough vram available, falling back to CPU only log

<!-- gh-comment-id:1970562656 --> @tylinux commented on GitHub (Feb 29, 2024): Maybe vram is not enough to load model, run `OLLAMA_DEBUG=1 ollama serve`, than run your model, see if there have `not enough vram available, falling back to CPU only` log
Author
Owner

@realcarlos commented on GitHub (Feb 29, 2024):

Thanks , I restarted ollama , it works now.

<!-- gh-comment-id:1970647957 --> @realcarlos commented on GitHub (Feb 29, 2024): Thanks , I restarted ollama , it works now.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#1717