Please review: Optimize vector multiply on X86

[Jakub Staszak]

Hi,

You should check whether N1C isn't NULL here:

+    N1C = dyn_cast<ConstantSDNode>(N1.getOperand(0));
+    // fold (mul x, 1) -> x
+    if (N1C->getAPIntValue() == 1)


Best,
Kuba

On Jun 16, 2013, at 2:22 PM, "Demikhovsky, Elena" <elena.demikhovsky at intel.com> wrote:

> <mul.diff>