[llvm] [AMDGPU][True16][CodeGen] add a 16bit d16 predicate for true16 mode (PR #156574)

[Joe Nash via llvm-commits]

================
@@ -1378,13 +1378,19 @@ let SubtargetPredicate = HasVmemPrefInsts in {
 }
 
 //===----------------------------------------------------------------------===//
-// Flat Patterns
+// Utilities
 //===----------------------------------------------------------------------===//
+class Mem_wrap<dag op, bit true16> {
----------------
Sisyph wrote:

I would probably call this class 'if_lo16' or 'extract_lo16'. The name Mem_wrap is too generic. This is a nice helper though, it simplifies this a lot. 

https://github.com/llvm/llvm-project/pull/156574