AMDGPU wave reduce intrinsics should support i16 and half values
Open
backend:AMDGPU
- Dominant language
- LLVM
- Stars
- 40.5k
- Forks
- 18.7k
- PR merge metrics
- PR metrics pending
Description
The `llvm.amdgcn.wave.reduce.` intrinsics currently support i32, i64, float and double. They should also support the same operations on i16 and half.
Contributor guide
Assessment
This issue has not been assessed yet.