[llvm] [AMDGPU] Eliminate unnecessary packing in wider f16 vectors for sdwa/opsel-able instruction (PR #137137)

[Pierre van Houtryve via llvm-commits]

================
@@ -1361,6 +1379,499 @@ bool SIPeepholeSDWALegacy::runOnMachineFunction(MachineFunction &MF) {
   return SIPeepholeSDWA().run(MF);
 }
 
+static bool isSrcDestFP16Bits(MachineInstr *MI, const SIInstrInfo *TII) {
----------------
Pierre-vh wrote:

@vg0204 Please use replies instead of editing the original comment. I got very confused while reading this.



https://github.com/llvm/llvm-project/pull/137137