[Mlir-commits] [clang] [llvm] [mlir] [openmp] [OpenMP][offload] Cross-team reductions with variable number of teams (PR #195102)

[Dhruva Chakrabarti]

dhruvachak wrote:

Can you split up this PR into smaller ones? If you are seeing performance improvements from better register usage, that is likely because of the inline changes. So your first PR can be just those inline changes on existing functions. Can you try that and quantify the performance improvement?

https://github.com/llvm/llvm-project/pull/195102