[Mlir-commits] [mlir] [MLIR][ROCDL] Add Scale Convert f8 <-> F32 Support for GFX950 (PR #125564)

[Corbin Robeck]

CRobeck wrote:

The LLVM IR tests should be place already for the backend builtin Ops that have already been added. See
[llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.scalef32.pk.gfx950.ll](https://github.com/llvm/llvm-project/blob/main/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.scalef32.pk.gfx950.ll)
for example.

https://github.com/llvm/llvm-project/pull/125564