Skip to content

Conversation

liqiangxl
Copy link
Collaborator

This limitation was already solved.

@liqiangxl liqiangxl requested a review from naoyam March 6, 2023 17:39
Copy link
Collaborator

@naoyam naoyam left a comment

Choose a reason for hiding this comment

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

Thanks for the cleanup.

Looks like there's still a performance issue as predicated blocks would also need to synchronize, but that seems like a minor problem.

@@ -4302,7 +4302,7 @@ TEST_F(NVFuserTest, FusionMultiGridReduction2_CUDA) {

FusionExecutor fe;
// NOLINTNEXTLINE(cppcoreguidelines-avoid-goto,hicpp-avoid-goto)
ASSERT_ANY_THROW(fe.compileFusion(&fusion));
ASSERT_NO_THROW(fe.compileFusion(&fusion));
Copy link
Collaborator

Choose a reason for hiding this comment

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

Can you add the validation of the results?

@liqiangxl liqiangxl force-pushed the llu/allow_pred_block branch from 2981c3e to 35b76f3 Compare March 7, 2023 14:11
@liqiangxl liqiangxl merged commit d5f10e1 into devel Mar 7, 2023
@liqiangxl liqiangxl deleted the llu/allow_pred_block branch March 7, 2023 16:34
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.

2 participants