<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<br>
<div class="moz-cite-prefix">On 11/20/2014 12:07 PM, Mark Heffernan
wrote:<br>
</div>
<blockquote
cite="mid:CAOi2v+RoUAvUvATS9EnVsQOQq3MQqtkOX0RqYx=GWXg7wwb7aA@mail.gmail.com"
type="cite">
<div dir="ltr">Rolling back this CL which just improves analysis
precision (trip count computation specifically) is probably not
the right approach.</div>
</blockquote>
Strongly, strongly agreed with this point. <br>
<blockquote
cite="mid:CAOi2v+RoUAvUvATS9EnVsQOQq3MQqtkOX0RqYx=GWXg7wwb7aA@mail.gmail.com"
type="cite">
<div dir="ltr">
<div>
<div><br>
</div>
<div>Mark</div>
</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Thu, Nov 20, 2014 at 11:13 AM, Hans
Wennborg <span dir="ltr"><<a moz-do-not-send="true"
href="mailto:hans@chromium.org" target="_blank">hans@chromium.org</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">Sorry for
reviving an old thread.<br>
<span class=""><br>
On Fri, Oct 10, 2014 at 10:39 AM, Mark Heffernan <<a
moz-do-not-send="true" href="mailto:meheff@google.com">meheff@google.com</a>>
wrote:<br>
> Author: meheff<br>
> Date: Fri Oct 10 12:39:11 2014<br>
> New Revision: 219517<br>
><br>
> URL: <a moz-do-not-send="true"
href="http://llvm.org/viewvc/llvm-project?rev=219517&view=rev"
target="_blank">http://llvm.org/viewvc/llvm-project?rev=219517&view=rev</a><br>
> Log:<br>
> This patch de-pessimizes the calculation of loop trip
counts in<br>
> ScalarEvolution in the presence of multiple exits.
Previously all<br>
> loops exits had to have identical counts for a loop
trip count to be<br>
> considered computable. This pessimization was
implemented by calling<br>
> getBackedgeTakenCount(L) rather than getExitCount(L,
ExitingBlock)<br>
> inside of
ScalarEvolution::getSmallConstantTripCount() (see the
FIXME<br>
> in the comments of that function). The pessimization
was added to fix<br>
> a corner case involving undefined behavior
(pr/16130). This patch more<br>
> precisely handles the undefined behavior case
allowing the pessimization<br>
> to be removed.<br>
<br>
</span>This caused a 50 k regression in Chromium's binary
size due to more<br>
unrolling. I suspect it's something we'll have to take, but
was<br>
wondering if you have any numbers showing performance
benefit here?<br>
<br>
Thanks,<br>
Hans<br>
</blockquote>
</div>
<br>
</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
llvm-commits mailing list
<a class="moz-txt-link-abbreviated" href="mailto:llvm-commits@cs.uiuc.edu">llvm-commits@cs.uiuc.edu</a>
<a class="moz-txt-link-freetext" href="http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits">http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits</a>
</pre>
</blockquote>
<br>
</body>
</html>