[PATCH] D14573: [X86][AVX512CD] add mask broadcast intrinsics

[Elena Demikhovsky via llvm-commits]

delena added inline comments.

================
Comment at: ../llvm/lib/Target/X86/Disassembler/X86Disassembler.cpp:836
@@ -835,2 +835,3 @@
   case TYPE_VK1:
+  case TYPE_VK4:
   case TYPE_VK8:
----------------
add VK2, VK32 and  VK64, please

================
Comment at: ../llvm/lib/Target/X86/X86ISelLowering.cpp:16431
@@ +16430,3 @@
+      MVT VT = Op.getSimpleValueType();
+      MVT MaskVT = MVT::getVectorVT(MVT::i1, VT.getVectorNumElements());
+      if (MaskVT.bitsGT(Mask.getSimpleValueType())) {
----------------
I think that MaskVT does not depend on VT. You just take 8 or 16 bits as one integer and broadcast them to all vector elements.

================
Comment at: ../llvm/lib/Target/X86/X86ISelLowering.cpp:16436
@@ +16435,3 @@
+          DAG.getNode(ISD::ANY_EXTEND, dl, newMaskVT, Mask));
+      }
+      else {
----------------
"else" one line up, like
} else {


http://reviews.llvm.org/D14573