[PATCH] D13632: AVX512 : i8/i16 vector CTLZ/CTLZ_ZERO_UNDEF lowering

[Elena Demikhovsky via llvm-commits]

delena added inline comments.

================
Comment at: lib/Target/X86/X86ISelLowering.cpp:1528
@@ -1519,3 +1527,3 @@
     }
     if (Subtarget->hasVLX() && Subtarget->hasCDI()) {
       setOperationAction(ISD::CTLZ,             MVT::v4i64, Legal);
----------------
All these types may be "Custom" for AVX-512F and extended to 512-bit vectors.

================
Comment at: lib/Target/X86/X86ISelLowering.cpp:17392
@@ +17391,3 @@
+
+  if (512 < (NumElems * 32)) {
+    // Split vector, it's L0 and Hi parts will be handled in next iteration.
----------------
NumElts >= 32 or NumElts > 16


Repository:
  rL LLVM

http://reviews.llvm.org/D13632