[Mlir-commits] [mlir] [ROCDL] Add the global.atomic.fadd intrinsic in ROCDL (PR #94486)

[Giuseppe Rossini]

giuseros wrote:

Ok, given I am exactly after that vector instruction, how about we merge this PR and then we enable vector support for `atomicrmw` once it emits the vector instruction?

https://github.com/llvm/llvm-project/pull/94486