[llvm] [AMDGPU] Recognize uniform expressions on tidx (PR #178533)

Stanislav Mekhanoshin via llvm-commits llvm-commits at lists.llvm.org
Wed Jan 28 15:59:02 PST 2026


rampitec wrote:

> This isn't correct in general, consider the case where the workgroup grid has 65 x 2 x 1

I thought we are not packing waves? I.e., the rest of the wave does not get lanes from a next workgroup.

https://github.com/llvm/llvm-project/pull/178533


More information about the llvm-commits mailing list