[GH-ISSUE #443] My CUDA Has seen the GPU but did not use it, why? #381

Open
opened 2026-10-31 21:14:31 +00:00 by bsakaguchi · 2 comments
Owner

Originally created by @Mark-ssss on GitHub (May 16, 2024).
Original GitHub issue: https://github.com/openai/CLIP/issues/443

it seems like: torch.cuda.device_count()=1, but torch.cuda.current_device()=0....
What I 've got:
NVIDIA GeForce GTX 1650 Ti
CUDA available: True (version:11.1)
cuDNN enabled: True(version:8.9.729)
pytorch 1.9.0
torchvision 0.10.0
prettytable
easydict

Originally created by @Mark-ssss on GitHub (May 16, 2024). Original GitHub issue: https://github.com/openai/CLIP/issues/443 it seems like: torch.cuda.device_count()=1, but torch.cuda.current_device()=0.... What I 've got: NVIDIA GeForce GTX 1650 Ti CUDA available: True (version:11.1) cuDNN enabled: True(version:8.9.729) pytorch 1.9.0 torchvision 0.10.0 prettytable easydict
Author
Owner

@Mark-ssss commented on GitHub (May 16, 2024):

QQ截图20240517014552
it comes out just like this orz....

<!-- gh-comment-id:2115853187 --> @Mark-ssss commented on GitHub (May 16, 2024): ![QQ截图20240517014552](https://github.com/openai/CLIP/assets/34942354/71984225-6289-4209-bf8d-244fb75875af) it comes out just like this orz....
Author
Owner

@cohml commented on GitHub (Oct 15, 2024):

Have you manually moved your tensors into VRAM?

model = model.to("cuda")
<!-- gh-comment-id:2414927341 --> @cohml commented on GitHub (Oct 15, 2024): Have you manually moved your tensors into VRAM? ```python model = model.to("cuda") ```
Sign in to join this conversation.
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
openai/CLIP#381
No description provided.