-
Notifications
You must be signed in to change notification settings - Fork 95
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
remove cuda-nvcc and document ptxas #398
Conversation
/condalock |
Should it be removed (user can install if needed) or just unpinned (user downgrades if needed)? Just wondering if unpinning is better, as some other things might be using it? |
Nothing else is using it except JAX and if someone is forcing XLA stuff in TensorFlow. In either case, I think a removal is warranted. It's not a good idea to mix channels anyway, and it is obviously causing a lot of confusion. So it is better left to the user to make their own calls while keeping these overarching images as simple as possible. |
From a cluster perspective, I'm happy to keep drivers up to date with whatever it is the cloud providers support (which is out of our hands). If we feel this isn't going to affect non-jax users, and they have a well documented workaround, I think this seems fine to me. Based on all this conversation, 2i2c now uses T4s as default GPUs for everyone 2i2c-org/infrastructure#1787. Thanks everyone! |
Co-authored-by: Wei Ji <[email protected]>
Closes #393
Fixes #390
cc @weiji14 @yuvipanda @dhruvbalwada @scottyhq