[PATCH] D38836: Use X86ISD::VBROADCAST in place of v2f64 X86ISD::MOVDDUP when AVX2 is available

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon accepted this revision.
RKSimon added a comment.
This revision is now accepted and ready to land.

LGTM



================
Comment at: test/CodeGen/X86/sse3-schedule.ll:566
   %3 = shufflevector <2 x double> %2, <2 x double> undef, <2 x i32> zeroinitializer
-  %4 = fadd <2 x double> %1, %3
+  %4 = fsub <2 x double> %3, %1 ; stop the movddup from being folded as a broadcast load
   ret <2 x double> %4
----------------
Make this change separately if you can. 


https://reviews.llvm.org/D38836