[llvm] r215702 - [x86] Add the initial skeleton of type-based dispatch for AVX vectors in
Chandler Carruth
chandlerc at gmail.com
Fri Aug 15 10:15:58 PDT 2014
On Fri, Aug 15, 2014 at 9:57 AM, Adam Nemet <anemet at apple.com> wrote:
> That’s a cool little function but it won't be usable in AVX512 context.
> There too we have 128-bit lane cross and non-crossing instructions, e.g.
> unpack is just an extension of AVX with the same intralane logic. We
> should probably gravitate toward formulating these as 128-bit crossing
> rather than half crossing.
I'm fine with that, but I suspect it should arrive when all of this is
generalized for AVX-512.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140815/02d23c53/attachment.html>
More information about the llvm-commits
mailing list