<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Sep 19, 2014 at 5:31 PM, 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:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Fri, Sep 19, 2014 at 1:10 PM, Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>> wrote:<br>
> The original vgrf splitting code was written assuming that with the<br>
> assumption that vgrfs came in two types: those that can be split into<br>
<br>
</span>s/ with the assumption that//<br>
<span class=""><br>
> single registers and those that can't be split at all<br>
<br>
</span>Period<br>
<span class=""><br>
> It was very<br>
> conservative and bailed as soon as more than one element of a register was<br>
> read or written. This won't work once we start allowing a regular MOV or<br>
> ADD operation to operate on multiple registers. This rewrite allows for<br>
> the case where a vgrf of size 5 may appropreately be split in to one<br>
<br>
</span>appropriately<br></blockquote><div><br></div><div>Typos fixed.<br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<span class=""><br>
> register of size 1 and two registers of size 2.<br>
<br>
</span>I'm not sure I understand enough yet to review.</blockquote></div></div></div>