[GH-ISSUE #2853] not work in 1080ti gpu #48251

Closed
opened 2026-04-28 07:23:04 -05:00 by GiteaMirror · 6 comments
Owner

Originally created by @basakamars on GitHub (Mar 1, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2853

Originally assigned to: @dhiltgen on GitHub.

+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 551.61 Driver Version: 551.61 CUDA Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce GTX 1080 Ti WDDM | 00000000:04:00.0 On | N/A |
| 30% 23C P2 56W / 250W | 286MiB / 11264MiB | 2% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|

not work in gpu, only run to cpu, why?

Originally created by @basakamars on GitHub (Mar 1, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2853 Originally assigned to: @dhiltgen on GitHub. +-----------------------------------------------------------------------------------------+ | NVIDIA-SMI 551.61 Driver Version: 551.61 CUDA Version: 12.4 | |-----------------------------------------+------------------------+----------------------+ | GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+========================+======================| | 0 NVIDIA GeForce GTX 1080 Ti WDDM | 00000000:04:00.0 On | N/A | | 30% 23C P2 56W / 250W | 286MiB / 11264MiB | 2% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ +-----------------------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=========================================================================================| **not work in gpu, only run to cpu, why?**
Author
Owner

@dhiltgen commented on GitHub (Mar 1, 2024):

Can you share the server.log?

<!-- gh-comment-id:1973830059 --> @dhiltgen commented on GitHub (Mar 1, 2024): Can you share the server.log?
Author
Owner

@schueszy commented on GitHub (Mar 4, 2024):

@basakamars
Which processor are you using?

<!-- gh-comment-id:1976946896 --> @schueszy commented on GitHub (Mar 4, 2024): @basakamars Which processor are you using?
Author
Owner

@basakamars commented on GitHub (Mar 7, 2024):

@schueszy
My processor is Intel Xeon X5675.

<!-- gh-comment-id:1983716693 --> @basakamars commented on GitHub (Mar 7, 2024): @schueszy My processor is Intel Xeon X5675.
Author
Owner

@dhiltgen commented on GitHub (Mar 7, 2024):

I believe this CPU lacks AVX, making this a dup of #2187

<!-- gh-comment-id:1984113561 --> @dhiltgen commented on GitHub (Mar 7, 2024): I believe this CPU lacks AVX, making this a dup of #2187
Author
Owner

@basakamars commented on GitHub (Mar 8, 2024):

Thank you! You're right processor is not support AVX.

<!-- gh-comment-id:1984904454 --> @basakamars commented on GitHub (Mar 8, 2024): Thank you! You're right processor is not support AVX.
Author
Owner

@dhiltgen commented on GitHub (Mar 8, 2024):

If possible, I'd recommend trying to use a CPU that does have at least AVX unless you know your model will completely fit into the GPU as we see ~4x improvement in speed between no-AVX and AVX.

<!-- gh-comment-id:1986441357 --> @dhiltgen commented on GitHub (Mar 8, 2024): If possible, I'd recommend trying to use a CPU that does have at least AVX unless you know your model will completely fit into the GPU as we see ~4x improvement in speed between no-AVX and AVX.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#48251