Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Releases? #1

Open
King-of-Infinite-Space opened this issue Aug 7, 2022 · 4 comments
Open

Releases? #1

King-of-Infinite-Space opened this issue Aug 7, 2022 · 4 comments

Comments

@King-of-Infinite-Space
Copy link

Thanks for providing the build for Navi10. I want to try it on my 5700XT but don't have the expertise to build everything myself. The releases in this repo seem to have been removed. Could you upload them again?

@xuhuisheng
Copy link
Owner

seems itis no need to rebuild ROCm for navi10, you can use

´´´
export HSA_OVERRIDE_GFX_VERSION=10.3.0
´´´

then we can pretend there is a navi21 card, at least mninst can run properly.

@King-of-Infinite-Space
Copy link
Author

Thanks. I installed ROCm 5.2.0 and tensorflow-rocm on Ubuntu 22.04. With this trick, models can run on 5700XT. However, it sometimes crashes due to memory allocation error. For now, I'm ok with running models on CPU.

@xuhuisheng
Copy link
Owner

could you show me how to reproduce the problem?

@King-of-Infinite-Space
Copy link
Author

It seems the problem is not from ROCm.
I was running tensorflow in jupyter notebook in VSCode. Sometimes the python process is not terminated after closing (microsoft/vscode-jupyter#11018), which does not release the VRAM. Therefore I get memory allocation error because VRAM is full.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants