LLVM GPU Working Group Meeting – Friday, December 2, 2022

Unfortunately, I am not available this Friday evening, IST. The intention is to attract attention to the review since we are not seeing any feedback. The ask remains the same … we need people familiar with divergence to go over the spec and the implementation.

From the ping I sent on discourse, and the review:

The concepts are sound, as far as they were reviewed by people working directly on that change. The analysis is new and does not affect existing pipelines. The AMDGPU pipeline is eager to use this analysis and is committed to maintaining it. Is this sufficient reason to let the patch through?

For submitting, at least a code review is necessary, which does not need domain expertise.

1 Like