Skip to content

amd bf16 gpu float support #2859

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.

Sign up for GitHub

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

Jump to bottom
Open
wants to merge 1 commit into
base: r2.18-rocm-enhanced
Choose a base branch
from
Open

amd bf16 gpu float support #2859

wants to merge 1 commit into from
+12 −4

Conversation

i-chaochen
Copy link
Collaborator

No description provided.

draganmladjenovic
draganmladjenovic requested changes Feb 24, 2025
View reviewed changes
Copy link

@draganmladjenovic draganmladjenovic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should probably not do this.

@i-chaochen
Copy link
Collaborator Author

We should probably not do this.

any reason why?

@pemeliya
Copy link

To better align MI300 and H100 HLO dumps, I have enabled these bf16 arithm ops on 0.5.0 branch to test this big HLO attached to: https://ontrack-internal.amd.com/browse/SWDEV-515395
But it looks like, it ran a bit slower then with ordinary f32 conversions.

@i-chaochen
Copy link
Collaborator Author

Yes, actually from the customer workloads, this makes performance worse..

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@draganmladjenovic draganmladjenovic draganmladjenovic requested changes

@pemeliya pemeliya Awaiting requested review from pemeliya

Requested changes must be addressed to merge this pull request.

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@i-chaochen @pemeliya @draganmladjenovic