[PATCH] D124385: AMDGPU: Special case divergence analysis for wave ID computation

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

In D124385#3484855 <https://reviews.llvm.org/D124385#3484855>, @nhaehnle wrote:

> I have a bunch of questions about this.

I've come back to this several times and been confused each time. I did write a program that convinced me that this is correct, and doesn't need to consider partial workgroups.



================
Comment at: llvm/test/Analysis/DivergenceAnalysis/AMDGPU/wave-id-computation.ll:27
+  %i14 = add i32 %i12, %i13
+  %i15 = sdiv exact i32 %i14, 32
+  ret i32 %i15
----------------
nhaehnle wrote:
> Why is this division exact?
that's what comes out of clang


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124385/new/

https://reviews.llvm.org/D124385