[PATCH] [x86] eliminate unnecessary shuffling/moves with unary scalar math ops (PR21507)

[Andrea Di Biagio]

================
Comment at: lib/Target/X86/X86InstrSSE.td:3417-3420
@@ -3417,4 +3416,6 @@
+
    def : Pat<(Intr VR128:$src),
              (vt (COPY_TO_REGCLASS(
              !cast<Instruction>("V"#NAME#Suffix##r) (ScalarVT (IMPLICIT_DEF)),
                     (ScalarVT (COPY_TO_REGCLASS VR128:$src, RC))), VR128))>;
+
----------------
Hi Sanjay,

do we still need this pattern? I maybe wrong, but your new pattern (at line 3413) should make this one dead.

http://reviews.llvm.org/D9504

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/