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

adjust to adapt ROCm5.4 #30

Open
wants to merge 1 commit into
base: performance
Choose a base branch
from
Open

Conversation

liligwu
Copy link
Collaborator

@liligwu liligwu commented Oct 26, 2022

Change the two intrinsic names to adapt ROCm5.4. Build and tests passed.
20221024_ROCm54_origin-performance_test_log.txt

Copy link

@HaiShaw HaiShaw left a comment

Choose a reason for hiding this comment

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

Please make this version specific.

@liligwu
Copy link
Collaborator Author

liligwu commented Oct 26, 2022

Please make this version specific.

It's not version specific. The build and tests pass on both ROCm5.4 and 5.3. Forgot to mention it, sorry.

@HaiShaw
Copy link

HaiShaw commented Oct 26, 2022

Discussed with @whchung

We should NOT have to change any app code like llvm.amdgcn.raw.buffer.load.f16 to llvm.amdgcn.raw.buffer.load.i16 cross ROCm releases.

The complain may be caused by HIP SDK, frontend, etc.

Let's check the LLVM IR on both versions with "-save-temps"

Copy link

@carlushuang carlushuang left a comment

Choose a reason for hiding this comment

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

LGTM

@liligwu
Copy link
Collaborator Author

liligwu commented Sep 24, 2024

This change is needed when switching to rocm5.4.
The ticket against compiler team was rejected https://ontrack-internal.amd.com/browse/SWDEV-364100

abojarov pushed a commit to abojarov/FBGEMM that referenced this pull request Oct 15, 2024
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

Successfully merging this pull request may close these issues.

4 participants