added ret and no-ret opt tests for floating point

atomic ops with double type.

Author: pravinjagtap

llvm/lib/Target/AMDGPU/AMDGPUAtomicOptimizer.cpp

@@ -921,8 +921,10 @@ void AMDGPUAtomicOptimizerImpl::optimizeAtomic(Instruction &I,
     Value *BroadcastI = nullptr;
 
     if (TyBitWidth == 64) {
-      Value *const ExtractLo = B.CreateTrunc(PHI, Int32Ty);
-      Value *const ExtractHi = B.CreateTrunc(B.CreateLShr(PHI, 32), Int32Ty);
+      Value *CastedPhi = B.CreateBitCast(PHI, IntNTy);
+      Value *const ExtractLo = B.CreateTrunc(CastedPhi, Int32Ty);
+      Value *const ExtractHi =
+          B.CreateTrunc(B.CreateLShr(CastedPhi, 32), Int32Ty);
       CallInst *const ReadFirstLaneLo =
           B.CreateIntrinsic(Intrinsic::amdgcn_readfirstlane, {}, ExtractLo);
       CallInst *const ReadFirstLaneHi =