[PATCH] D26862: [AMDGPU] Change frexp.exp intrinsic to return i16 for f16 input

[Matt Arsenault via llvm-commits]

arsenm added inline comments.


================
Comment at: test/CodeGen/AMDGPU/llvm.amdgcn.frexp.exp.f16.ll:17
   ret void
 }
----------------
Can you add additional tests where the result is sext and zext to i32 to make sure it is properly sign extended


https://reviews.llvm.org/D26862