<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Thu, May 31, 2018 at 10:24 AM, Matt Turner <span dir="ltr"><<a href="mailto:mattst88@gmail.com" target="_blank">mattst88@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><span class="gmail-">On Thu, May 24, 2018 at 2:56 PM, Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>> wrote:<br>
> From: Francisco Jerez <<a href="mailto:currojerez@riseup.net">currojerez@riseup.net</a>><br>
><br>
> Reviewed-by: Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>><br>
<br>
</span>Some small explanation that this is possible since we're able to rely<br>
on lower_simd_width() to do the hard work for us might be nice.<br>
</blockquote></div></div><div class="gmail_extra"><br></div><div class="gmail_extra"> The original code manually handled splitting the MOVs to 8-wide to<br> handle various regioning restrictions. Now that we have a SIMD width<br> splitting pass that handles these things, we can just emit everything at<br> the full width and let the SIMD splitting pass handle it. We also now<br> have a useful "subscript" helper which is designed exactly for the case<br> where you want to take a W type and read it as a vector of Bs so we may<br> as well use that too.<br><br><br></div></div>