<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Nov 12, 2015 at 11:55 PM, Owen Anderson <span dir="ltr"><<a href="mailto:resistor@mac.com" target="_blank">resistor@mac.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">resistor added a comment.<br>
<span class=""><br>
In <a href="http://reviews.llvm.org/D12345#288471" rel="noreferrer" target="_blank">http://reviews.llvm.org/D12345#288471</a>, @dberlin wrote:<br>
<br>
> The last status i saw was: "I'll try seeing what little needs to be done to<br>
> converge in the couple cases that I have (one above) and a superficial look<br>
> at why the second reassociate improves codegen."<br>
><br>
> I saw no update on that ;-)<br>
<br>
<br>
</span>Would it be reasonable to stage that investigation? AFAICT the changes already proposed (with Chad's comments incorporated) are a strict improvement, and I at least am seeing pretty severe performance regressions from their absence. Would you be alright with going ahead and landing those?<br></blockquote><div><br></div><div><br></div><div>I'm not opposed, I'd like to see compile time numbers first since it transforms reassociate into an iterative algorithm.</div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">For my use case, at least, N-ary reassociation is not really appropriate as I also heavily depend on reassociation of floating point arithmetic. </blockquote><div><br></div><div>I'm not sure why these two statements go together ;-)</div><div>What am i missing?</div><div> </div><div>The difference is one of a global vs local algorithm, not of floating point vs integer (n-ary reassociate is basically just a global version of reassociation that makes choices with more context)</div><div><br></div><div>You want to move towards the global algorithm no matter what you do.</div><div><br></div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> So, I'm stuck with local reassociation, and the problem described here is making today's LLVM integer factors worse than last year's for me.<br></blockquote><div><br></div><div>I understand and am sympathetic to the second part. Still trying to understand the first, so if you could lay it out for me, i'd appreciate it ;-)</div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="HOEnZb"><div class="h5"><br>
<br>
Repository:<br>
rL LLVM<br>
<br>
<a href="http://reviews.llvm.org/D12345" rel="noreferrer" target="_blank">http://reviews.llvm.org/D12345</a><br>
<br>
<br>
<br>
</div></div></blockquote></div><br></div></div>