[clang] [llvm] [AMDGPU] Adding the amdgpu-num-work-groups function attribute (PR #79035)

[Matt Arsenault via llvm-commits]

arsenm wrote:

> If that’s the case, is 1 the best-case scenario?

Yes

> @krzysz00 Do you suggest that the actual value doesn't matter? Like, 128 is not better or worse than 256.

This enables known bits optimizations. Less is always better, but the benefits can be marginal. 



https://github.com/llvm/llvm-project/pull/79035