[llvm] [AMDGPU] Add CodeGen support for GFX12 s_mul_u64 (PR #75825)

[Matt Arsenault via llvm-commits]

================
@@ -419,6 +427,32 @@ void AMDGPUPostLegalizerCombinerImpl::applyCombineSignExtendInReg(
   MI.eraseFromParent();
 }
 
+bool AMDGPUPostLegalizerCombinerImpl::matchCombine_s_mul_u64(
+    MachineInstr &MI, unsigned &NewOpcode) const {
+  Register Src0 = MI.getOperand(1).getReg();
+  Register Src1 = MI.getOperand(2).getReg();
+  if (MRI.getType(Src0) != LLT::scalar(64))
+    return false;
+
+  if (KB->getKnownBits(Src0).countMinLeadingZeros() >= 32 &&
+      KB->getKnownBits(Src1).countMinLeadingZeros() >= 32) {
+    NewOpcode = AMDGPU::G_AMDGPU_S_MUL_U64_U32;
+    return true;
+  }
+
+  if (KB->computeNumSignBits(Src0) >= 33 &&
+      KB->computeNumSignBits(Src1) >= 33) {
----------------
arsenm wrote:

Try Src1 first 

https://github.com/llvm/llvm-project/pull/75825