[X86] AMD Zen 3 has fast variable per-lane shuffles

... but lane-crossing shuffles are slow.
llvm · Jun 1, 2021 · a3b8695 · a3b8695
1 parent cf9b1f7
commit a3b8695
Showing 1 changed file with 3 additions and 1 deletion.
diff --git a/llvm/lib/Target/X86/X86.td b/llvm/lib/Target/X86/X86.td
@@ -1111,7 +1111,9 @@ def ProcessorFeatures {
                                                   FeaturePKU,
                                                   FeatureVAES,
                                                   FeatureVPCLMULQDQ];
-  list<SubtargetFeature> ZN3AdditionalTuning = [FeatureMacroFusion];
+  list<SubtargetFeature> ZN3AdditionalTuning =
+    [FeatureMacroFusion,
+     FeatureFastVariablePerLaneShuffle];
   list<SubtargetFeature> ZN3Tuning =
     !listconcat(ZNTuning, ZN3AdditionalTuning);
   list<SubtargetFeature> ZN3Features =