[llvm] AMDGPU: Use pattern to select instruction for intrinsic llvm.fptrunc.round (PR #105761)

[Matt Arsenault via llvm-commits]

================
@@ -343,7 +343,7 @@ HANDLE_TARGET_OPCODE(G_FREEZE)
 HANDLE_TARGET_OPCODE(G_CONSTANT_FOLD_BARRIER)
 
 // INTRINSIC fptrunc_round intrinsic.
-HANDLE_TARGET_OPCODE(G_INTRINSIC_FPTRUNC_ROUND)
+HANDLE_TARGET_OPCODE(G_FPTRUNC_ROUND)
----------------
arsenm wrote:

Rename should be done separately. It is not necessary to move this to a pattern 

https://github.com/llvm/llvm-project/pull/105761