[PATCH] D80095: AMDGPU/GlobalISel: Fix splitting wide VALU, non-vector loads

[Austin Kerbow via Phabricator via llvm-commits]

kerbowa accepted this revision.
kerbowa added a comment.
This revision is now accepted and ready to land.

LGTM



================
Comment at: llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-load.mir:44
+    %tmp0 = call i32 @llvm.amdgcn.workitem.id.x() #0
+    %constant.not.uniform = getelementptr i256, i256 addrspace(4)* %in, i32 %tmp0
+    %tmp2 = load i256, i256 addrspace(4)* %constant.not.uniform
----------------
Random nit, the suffix for constant.not.uniform is not consistent with the other tests.


================
Comment at: llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-load.mir:51
+    %tmp0 = call i32 @llvm.amdgcn.workitem.id.x() #0
+    %constant.not.uniform = getelementptr <16 x i16>, <16 x i16> addrspace(4)* %in, i32 %tmp0
+    %tmp2 = load <16 x i16>, <16 x i16> addrspace(4)* %constant.not.uniform
----------------
Ditto.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D80095/new/

https://reviews.llvm.org/D80095