[llvm-branch-commits] [llvm] AMDGPU/GlobalISel: Switch RegBankLegalize lowering to extended LLTs (PR #208270)

[Chinmay Deshpande via llvm-branch-commits]

================
@@ -933,7 +936,7 @@ bool RegBankLegalizeHelper::lowerSplitTo32Select(MachineInstr &MI) {
   LLT DstTy = MRI.getType(Dst);
   assert(DstTy == V4S16 || DstTy == V2S32 || DstTy == S64 ||
          (DstTy.isPointer() && DstTy.getSizeInBits() == 64));
-  LLT Ty = DstTy == V4S16 ? V2S16 : S32;
+  LLT Ty = DstTy.isFloat() ? S32 : DstTy.divide(2);
----------------
chinmaydd wrote:

Is this check needed ? Looking at `lowerSplitTo32` (and the asserts) I dont think we need to do the `isFloat` check 

https://github.com/llvm/llvm-project/pull/208270