[llvm] X86: make VBMI2 funnel shifts use VSHLD/VSHRD for const splats (PR #169401)

[via llvm-commits]

================
@@ -57622,6 +57626,49 @@ static SDValue combineFP_TO_xINT_SAT(SDNode *N, SelectionDAG &DAG,
   return SDValue();
 }
 
+// Combiner: turn uniform-constant splat funnel shifts into VSHLD/VSHRD
----------------
ArnavM3434 wrote:

I believe it is because in the combine we can easily check for constant splats and also type legality. I'm not sure how we'd do this with TableGen patterns. @RKSimon do you agree?

https://github.com/llvm/llvm-project/pull/169401