[llvm] [AMDGPU] Add support for llvm.lround and llvm.lrint intrinsics lowering. (PR #96817)

[Matt Arsenault via llvm-commits]

================
@@ -5928,6 +5928,27 @@ void Verifier::visitIntrinsicCall(Intrinsic::ID ID, CallBase &Call) {
     }
     break;
   }
+  case Intrinsic::lround:
+  case Intrinsic::llround: {
+    Type *ValTy = Call.getArgOperand(0)->getType();
----------------
arsenm wrote:

This also needs dedicated verifier tests 

https://github.com/llvm/llvm-project/pull/96817