[llvm] [X86][CodeGen] Support EVEX compression: NDD to nonNDD (PR #77731)

Phoebe Wang via llvm-commits llvm-commits at lists.llvm.org
Thu Jan 11 05:58:31 PST 2024


================
@@ -220,6 +220,27 @@ static bool performCustomAdjustments(MachineInstr &MI, unsigned NewOpc) {
   return true;
 }
 
+static bool isRedundantNewDataDest(MachineInstr &MI, const X86Subtarget &ST) {
+  // add rbx, rbx, rax / add rbx, rax, rbx -> add rbx, rax
----------------
phoebewang wrote:

It would be better if you use a real node, e.g., ADDrr_ND instead of a add.

https://github.com/llvm/llvm-project/pull/77731


More information about the llvm-commits mailing list