<?xml version="1.0"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
	<id>http://mw.hh.se/wg211/index.php?action=history&amp;feed=atom&amp;title=WG211%2FM12Scholz</id>
	<title>WG211/M12Scholz - Revision history</title>
	<link rel="self" type="application/atom+xml" href="http://mw.hh.se/wg211/index.php?action=history&amp;feed=atom&amp;title=WG211%2FM12Scholz"/>
	<link rel="alternate" type="text/html" href="http://mw.hh.se/wg211/index.php?title=WG211/M12Scholz&amp;action=history"/>
	<updated>2026-04-05T19:32:49Z</updated>
	<subtitle>Revision history for this page on the wiki</subtitle>
	<generator>MediaWiki 1.43.5</generator>
	<entry>
		<id>http://mw.hh.se/wg211/index.php?title=WG211/M12Scholz&amp;diff=749&amp;oldid=prev</id>
		<title>Ups: Created page with &quot;&#039;&#039;&#039;Data-Layout Inference for Generating Vectorised Code&#039;&#039;&#039; by Sven-Bodo Scholz  SIMD instructions of modern CPUs are crucially important for the performance of compute-intensive ...&quot;</title>
		<link rel="alternate" type="text/html" href="http://mw.hh.se/wg211/index.php?title=WG211/M12Scholz&amp;diff=749&amp;oldid=prev"/>
		<updated>2013-03-08T21:22:16Z</updated>

		<summary type="html">&lt;p&gt;Created page with &amp;quot;&amp;#039;&amp;#039;&amp;#039;Data-Layout Inference for Generating Vectorised Code&amp;#039;&amp;#039;&amp;#039; by Sven-Bodo Scholz  SIMD instructions of modern CPUs are crucially important for the performance of compute-intensive ...&amp;quot;&lt;/p&gt;
&lt;p&gt;&lt;b&gt;New page&lt;/b&gt;&lt;/p&gt;&lt;div&gt;&amp;#039;&amp;#039;&amp;#039;Data-Layout Inference for Generating Vectorised Code&amp;#039;&amp;#039;&amp;#039; by Sven-Bodo Scholz&lt;br /&gt;
&lt;br /&gt;
SIMD instructions of modern CPUs are crucially important for the&lt;br /&gt;
performance of compute-intensive programs. Often, auto-vectorisation leads&lt;br /&gt;
to sub-optimal results due to the data-layout chosen by the application&lt;br /&gt;
programmer.  The reasons for choosing an unfavourable layout can be&lt;br /&gt;
manifold: software engineering principles may contradict performance&lt;br /&gt;
demands, the layout demands may depend on the target architecture a program&lt;br /&gt;
was first run on, etc. However, most of the time it is simply a lack of&lt;br /&gt;
awareness of the programmer.&lt;br /&gt;
&lt;br /&gt;
In this talk, we present an approach towards liberating the programmer from &lt;br /&gt;
the burden of considering the suitability of his data structures for&lt;br /&gt;
vectorisation. We present an inference scheme that allows compilers to&lt;br /&gt;
identify suitable data-layouts and a refactoring scheme to modify the code&lt;br /&gt;
accordingly.  We demonstrate our ideas by means of a running example and&lt;br /&gt;
present some very encouraging initial performance results of our technique.&lt;/div&gt;</summary>
		<author><name>Ups</name></author>
	</entry>
</feed>