[llvm] [AMDGPU][True16][MC] true16 for more VOP1 instructions (PR #108412)

[Ivan Kosarev via llvm-commits]

================
@@ -939,6 +923,14 @@ multiclass VOP1_Real_FULL_with_name_gfx11_gfx12<bits<9> op, string opName,
   VOP1_Real_FULL_with_name<GFX11Gen, op, opName, asmName>,
   VOP1_Real_FULL_with_name<GFX12Gen, op, opName, asmName>;
 
+multiclass VOP1_Real_FULL_t16_and_f16_gfx11_gfx12<bits<9> op, string asmName,
+                                          string opName = NAME> {
+  defm opName#"_t16" :
+	VOP1_Real_FULL_with_name_gfx11_gfx12<op, opName#"_t16", asmName>;
+  defm opName#"_fake16":
+	VOP1_Real_FULL_with_name_gfx11_gfx12<op, opName#"_fake16", asmName>;
+}
----------------
kosarev wrote:

Combining these looks great, thanks.

https://github.com/llvm/llvm-project/pull/108412