[llvm] AMDGPU: Set RegTupleAlignUnits on _Lo256_Align2 class (PR #159383)

[via llvm-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-amdgpu

Author: Matt Arsenault (arsenm)

<details>
<summary>Changes</summary>



---
Full diff: https://github.com/llvm/llvm-project/pull/159383.diff


1 Files Affected:

- (modified) llvm/lib/Target/AMDGPU/SIRegisterInfo.td (+3-1) 


``````````diff
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.td b/llvm/lib/Target/AMDGPU/SIRegisterInfo.td
index 4e1876db41d3d..8f1dd6244f20d 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.td
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.td
@@ -1025,7 +1025,9 @@ multiclass VRegClass<int numRegs, list<ValueType> regTypes, dag regList> {
 
     // Aligned register tuples starting with low 256 vgprs
     def _Lo256_Align2 : VRegClassBase<numRegs, regTypes,
-        (trunc (decimate regList, 2), !div(!sub(258, numRegs), 2))>;
+        (trunc (decimate regList, 2), !div(!sub(258, numRegs), 2))> {
+      let RegTupleAlignUnits = 2;
+     }
   }
 }
 

``````````

</details>


https://github.com/llvm/llvm-project/pull/159383