[llvm] [X86] Attempt to use VPMADD52L/VPMULUDQ instead of VPMULLQ on slow VPMULLQ targets (or when VPMULLQ is unavailable) (PR #171760)

[Simon Pilgrim via llvm-commits]

================
@@ -1049,14 +1052,11 @@ def ProcessorFeatures {
                                                   FeatureVBMI,
                                                   FeatureIFMA,
                                                   FeatureSHA];
-  list<SubtargetFeature> CNLTuning = [TuningFastGather,
-                                      TuningMacroFusion,
-                                      TuningSlow3OpsLEA,
-                                      TuningSlowDivide64,
+  list<SubtargetFeature> CNLTuning = [TuningFastGather, TuningMacroFusion,
+                                      TuningSlow3OpsLEA, TuningSlowDivide64,
----------------
RKSimon wrote:

(style) avoid 80col rules - just keep 1 flag per line to minimise changes

https://github.com/llvm/llvm-project/pull/171760