[llvm-commits] VPBLENDD match in AVX2 - please review

[Nadav Rotem]

On Dec 5, 2012, at 12:34 AM, "Demikhovsky, Elena" <elena.demikhovsky at intel.com> wrote:

> 1. Because I want to generate VPBLENDD for integers instead of VBLENDPS (I don't gain performance in this case, just in respect for somebody who invented VPBLENDD in AVX2)

I think that this is something that the domain fixer should handle. 

> 2. I can't convert i16 vectors to fp

Good point.  In that case, maybe all of the blends should be converted integer-blends ? 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20121205/ef747792/attachment.html>