[PATCH] D10683: AVX-512 vector shuffle lowering

[Simon Pilgrim via llvm-commits]

RKSimon added a comment.

I think its almost there - just a few questions that I missed earlier.


================
Comment at: lib/Target/X86/X86ISelLowering.cpp:4627
@@ -4623,1 +4626,3 @@
     return false;
+  case X86ISD::VPERMV: {
+    IsUnary = true;
----------------
All this code looks very similar to the X86ISD::PSHUFB implementation - is there anyway that they can be merged?

================
Comment at: lib/Target/X86/X86ISelLowering.cpp:4690
@@ +4689,3 @@
+  }
+  case X86ISD::VPERMV3: {
+    IsUnary = false;
----------------
All this code looks very similar to the X86ISD::PSHUFB implementation - is there anyway that they can be merged?

================
Comment at: lib/Target/X86/X86ISelLowering.cpp:10547
@@ +10546,3 @@
+  }
+  if (!Unpckl && !Unpckh)
+    return SDValue();
----------------
Can we take into account duplicated inputs here? What about matching the inputs if they were swapped?


Repository:
  rL LLVM

http://reviews.llvm.org/D10683