[PATCH] D83249: AMDGPU: Handle llvm.amdgcn.buffer.{load|store}.v2i16 intrinsics

[Matt Arsenault via Phabricator via llvm-commits]

arsenm requested changes to this revision.
arsenm added a comment.
This revision now requires changes to proceed.

Can you also make sure this works with globalisel? I think it should already, but it would be good to double check the tests are there



================
Comment at: llvm/test/CodeGen/AMDGPU/llvm.amdgcn.buffer.load-store.v2i16.ll:34-36
+  %4 = ptrtoint i16* %0 to i64
+  %5 = bitcast i64 %4 to <2 x i32>
+  %6 = shufflevector <2 x i32> %5, <2 x i32> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
----------------
You can remove most of this setup code (also tests should use named values).


================
Comment at: llvm/test/CodeGen/AMDGPU/llvm.amdgcn.buffer.load-store.v2i16.ll:40
+  %9 = shl i32 %8, 1
+  %10 = tail call <2 x i16> @llvm.amdgcn.buffer.load.v2i16(<4 x i32> %7, i32 0, i32 %9, i1 zeroext false, i1 zeroext false)
+  ret <2 x i16> %10
----------------
These are the legacy intrinsics. Only the .raw/.struct versions matter now. This should also add tests with <2 x half>


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D83249/new/

https://reviews.llvm.org/D83249