<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: QLogic Aims for the Fences with Infiniband Fabric Suite</title>
	<atom:link href="http://insidehpc.com/2010/05/19/qlogic-aims-for-the-fences-with-infiniband-fabric-suite/feed/" rel="self" type="application/rss+xml" />
	<link>http://insidehpc.com/2010/05/19/qlogic-aims-for-the-fences-with-infiniband-fabric-suite/</link>
	<description>HPC News Without the Noise for Supercomputing Professionals &#124; insideHPC</description>
	<lastBuildDate>Sun, 19 May 2013 10:46:54 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.1.1</generator>
	<item>
		<title>By: Gilad</title>
		<link>http://insidehpc.com/2010/05/19/qlogic-aims-for-the-fences-with-infiniband-fabric-suite/#comment-227982</link>
		<dc:creator>Gilad</dc:creator>
		<pubDate>Thu, 20 May 2010 16:19:11 +0000</pubDate>
		<guid isPermaLink="false">http://insidehpc.com/?p=11419#comment-227982</guid>
		<description>There are few missing points and wrong data. &quot;traditional &quot; InfiniBand does not peak at 7M messages per second at all. The numbers demonstrated are much much higher. Furthermore, the approach described here requires many CPU cycles to drive the network, so in real application works, you will see many issues. You can refer to - http://www.hpcwire.com/features/Network-based-Processing-Versus-Host-based-Processing-93961524.html for more info.

Second, queue-pair is the InfiniBand advantage. It allows creating 16M virtual NICs with full protection and isolation between them. But if you avoid that, you put a big load on the CPU. It becomes more like using TCP for HPC. Not very efficient. InfiniBand target was to provide the most efficient interconnect, and this is what other vendors provide.</description>
		<content:encoded><![CDATA[<p>There are few missing points and wrong data. &#8220;traditional &#8221; InfiniBand does not peak at 7M messages per second at all. The numbers demonstrated are much much higher. Furthermore, the approach described here requires many CPU cycles to drive the network, so in real application works, you will see many issues. You can refer to &#8211; <a href="http://www.hpcwire.com/features/Network-based-Processing-Versus-Host-based-Processing-93961524.html" rel="nofollow">http://www.hpcwire.com/features/Network-based-Processing-Versus-Host-based-Processing-93961524.html</a> for more info.</p>
<p>Second, queue-pair is the InfiniBand advantage. It allows creating 16M virtual NICs with full protection and isolation between them. But if you avoid that, you put a big load on the CPU. It becomes more like using TCP for HPC. Not very efficient. InfiniBand target was to provide the most efficient interconnect, and this is what other vendors provide.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
