-
Notifications
You must be signed in to change notification settings - Fork 210
Update XLA #1614
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
Update XLA #1614
Conversation
Co-authored-by: Paulo Valente <[email protected]>
@polvalente all good! |
Let me test this locally on CUDA later today before merging! |
Perfect! |
exla/c_src/exla/custom_calls/eigh.h
Outdated
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We already wanted to merge custom calls into the XLA project so we can decrease the compile time, let's take the opportunity and do it now?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I already released new XLA. The update took long this time, so I didn't want to add extra stuff to postpone it even longer. Let's do it separately once we have more time to prioritise this :)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
ok!
I've tried this PR with a 50xx card for a bit, and it looks good. |
@petrkozorezov fantastic, thank you for letting us know! |
Bumps XLA to recent revision.
We were still using the oldest API for defining custom calls and now a bunch of deprecation errors have been logged. I went ahead and updated to the latest typed FFI API, which also simplified the MLIR generation nicely.
Closes #1599.