[PATCH] Global Structure Vectorization

[Renato Golin]

On 20 February 2013 20:04, Arnold Schwaighofer <aschwaighofer at apple.com>wrote:

> Thanks for working on this. The code looks correct. I agree with Arnold
> that this function is becoming even more complicated and it would be nice
> to refactor it somehow, but we can do it later. I don't mind if the tests
> are in one file or multiple files.  I assume that you ran the test-suite
> and found no failures. Did you see any newly vectorized loops ?
>
> Since Renato seems to be at the pub right now :). I'll answer part of your
> question: no new failures.
>

Thanks, yes, no new failures, though I haven't inspected generated assembly
to search for newly-vectorized code. I'd also assume there is very little
at this stage, since this is a very specific case (that will grow into a
generic one).

I'll run this against a local Livermore Loops and see if it makes any
difference at this stage. I pretend to introduce it again, slowly, one
kernel at a time, making sure we get no undefined behaviour in the
initialization.

cheers,
--renato
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130220/949ec04c/attachment.html>