<div dir="ltr">Ok, will do, <div>I should remark that the current tests for vanil x86_32 and x86_64 ghc calling conventions are using check-next currently, so i was copying that convention. Should I update the patch to change those to check too?</div>
</div><div class="gmail_extra"><br><br><div class="gmail_quote">On Mon, Jul 8, 2013 at 4:54 PM, Stephen Lin <span dir="ltr"><<a href="mailto:swlin@post.harvard.edu" target="_blank">swlin@post.harvard.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi Carter,<br>
<br>
I'm not sure if you can really depend on loads being ordered in the<br>
order you specify them in the IR; even if they are now, they might not<br>
be in the future and it will add to the difficulty of testing changes<br>
later, so it's better not to specify things you don't actually depend<br>
upon.<br>
<br>
Can you change the loads to volatile loads (just put the keyword<br>
"volatile" after the load) and change the CHECK-NEXT lines to simply<br>
CHECK? I haven't tried it yet but think that would work and be more<br>
robust to future changes.<br>
<br>
Thanks,<br>
Stephen<br>
<div><div class="h5"><br>
On Mon, Jul 8, 2013 at 1:37 PM, Carter Tazio Schonwald<br>
<<a href="mailto:carter.schonwald@gmail.com">carter.schonwald@gmail.com</a>> wrote:<br>
> Hey All,<br>
><br>
> currently the GHC calling convention doesnt have support for using the AVX<br>
> 256bit width registers. Attached is a patch that augments the GHC x86-64<br>
> calling convention with that support when vector types of that size are<br>
> used.<br>
><br>
> This change has been Ok'd by the GHC HQ devs responsible for the SIMD<br>
> support recently added to ghc (as well as the principal author of the llvm<br>
> backend for ghc ). see the ghc ticket here<br>
> <a href="http://hackage.haskell.org/trac/ghc/ticket/8033" target="_blank">http://hackage.haskell.org/trac/ghc/ticket/8033</a> for their indications of<br>
> approval<br>
><br>
> it'd be really great to have this patch in both the 3.4 and the pending 3.3<br>
> point release, because then the next GHC release could get some additional<br>
> work to support AVX2 now rather than later. (in addition to the current<br>
> support for 128bit simd)<br>
><br>
> i've included the patch and an additional test case in the diff attached<br>
> below<br>
><br>
> thanks!<br>
> -Carter<br>
><br>
</div></div>> _______________________________________________<br>
> llvm-commits mailing list<br>
> <a href="mailto:llvm-commits@cs.uiuc.edu">llvm-commits@cs.uiuc.edu</a><br>
> <a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits" target="_blank">http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits</a><br>
><br>
</blockquote></div><br></div>