AMDGPU: Enable ExpandLargeFpConvert for > 64-bit types

Fixes casts between double/float/half and i128. The pass seems to be
broken for bfloat though. I also believe we could have a better implementation
which attempts to make use the native 32-bit conversion instructions like
the 64-bit expansion does.

Author: arsenm

llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp

@@ -582,6 +582,7 @@ AMDGPUTargetLowering::AMDGPUTargetLowering(const TargetMachine &TM,
 
   setMaxAtomicSizeInBitsSupported(64);
   setMaxDivRemBitWidthSupported(64);
+  setMaxLargeFPConvertBitWidthSupported(64);
 }
 
 bool AMDGPUTargetLowering::mayIgnoreSignedZero(SDValue Op) const {