<?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: Future Storage Systems: Part 6: GPU accelerated storage</title>
	<atom:link href="http://flickerdown.com/2008/11/future-storage-systems-part-6-gpu-accelerated-storage/feed/" rel="self" type="application/rss+xml" />
	<link>http://flickerdown.com/2008/11/future-storage-systems-part-6-gpu-accelerated-storage/</link>
	<description>a Blog dedicated to storage and technology</description>
	<lastBuildDate>Tue, 06 Jul 2010 21:18:22 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.0</generator>
	<item>
		<title>By: Dave Graham</title>
		<link>http://flickerdown.com/2008/11/future-storage-systems-part-6-gpu-accelerated-storage/comment-page-1/#comment-124</link>
		<dc:creator>Dave Graham</dc:creator>
		<pubDate>Wed, 12 Nov 2008 18:44:37 +0000</pubDate>
		<guid isPermaLink="false">http://flickerdown.com/?p=259#comment-124</guid>
		<description>Aryan,&lt;br&gt;&lt;br&gt;thanks for the comments.  When you look at the specialized workloads that GPGPUs can provide additional processing power, it&#039;s amazing that they haven&#039;t been integrated sooner.  I downloaded an interesting whitepaper entitled &quot;QP: A Heterogeneous Multi-Accelerator Cluster&quot; done by the University of Illinois at Urbana-Champaign that really highlighted the strengths of using both foundational CPUs (OS and programmatic interface) as well as specialized GPGPUs for determined workloads.  I think, based on what you&#039;ve written above, that the encryption (data security) vertical is well suited to using these devices.  From a storage perspective, encrypt/decrypt, deduplication (calculation and stream intensive), and even more complex parity workloads will benefit highly from GPGPUs.&lt;br&gt;&lt;br&gt;Thanks again!&lt;br&gt;&lt;br&gt;cheers,&lt;br&gt;&lt;br&gt;Dave Graham</description>
		<content:encoded><![CDATA[<p>Aryan,</p>
<p>thanks for the comments.  When you look at the specialized workloads that GPGPUs can provide additional processing power, it&#39;s amazing that they haven&#39;t been integrated sooner.  I downloaded an interesting whitepaper entitled &#8220;QP: A Heterogeneous Multi-Accelerator Cluster&#8221; done by the University of Illinois at Urbana-Champaign that really highlighted the strengths of using both foundational CPUs (OS and programmatic interface) as well as specialized GPGPUs for determined workloads.  I think, based on what you&#39;ve written above, that the encryption (data security) vertical is well suited to using these devices.  From a storage perspective, encrypt/decrypt, deduplication (calculation and stream intensive), and even more complex parity workloads will benefit highly from GPGPUs.</p>
<p>Thanks again!</p>
<p>cheers,</p>
<p>Dave Graham</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Aryan</title>
		<link>http://flickerdown.com/2008/11/future-storage-systems-part-6-gpu-accelerated-storage/comment-page-1/#comment-123</link>
		<dc:creator>Aryan</dc:creator>
		<pubDate>Wed, 12 Nov 2008 17:28:13 +0000</pubDate>
		<guid isPermaLink="false">http://flickerdown.com/?p=259#comment-123</guid>
		<description>I expect GPGPU functions to be integrated into both software and hardware.&lt;br&gt;Software at first specialised apps later into the OS, hardware more and more into southbridges/hybrid solution. Ye ole M$ is barking at the tree in this regard, but progress in this field has been so storming that GPGPU integration is only a small problem compared with the huge security risks that are caused by this massive increase in avaliable processing power; a workload that a CPU took years is reduced a weeks. Ie: bruteforcing strong MD5 hashes. [ now listening GSYBE - Static ] Imagine a marriage between rainbow and bruteforce at this level. It takes me seconds to bruteforce any MD5 hash of 8 or less characters/numbers....  If m$ KEEPS ON SLACKING  they&#039;ll also miss this train as it streams into the cloud.</description>
		<content:encoded><![CDATA[<p>I expect GPGPU functions to be integrated into both software and hardware.<br />Software at first specialised apps later into the OS, hardware more and more into southbridges/hybrid solution. Ye ole M$ is barking at the tree in this regard, but progress in this field has been so storming that GPGPU integration is only a small problem compared with the huge security risks that are caused by this massive increase in avaliable processing power; a workload that a CPU took years is reduced a weeks. Ie: bruteforcing strong MD5 hashes. [ now listening GSYBE - Static ] Imagine a marriage between rainbow and bruteforce at this level. It takes me seconds to bruteforce any MD5 hash of 8 or less characters/numbers&#8230;.  If m$ KEEPS ON SLACKING  they&#39;ll also miss this train as it streams into the cloud.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: dave_graham</title>
		<link>http://flickerdown.com/2008/11/future-storage-systems-part-6-gpu-accelerated-storage/comment-page-1/#comment-42</link>
		<dc:creator>dave_graham</dc:creator>
		<pubDate>Wed, 12 Nov 2008 12:44:37 +0000</pubDate>
		<guid isPermaLink="false">http://flickerdown.com/?p=259#comment-42</guid>
		<description>Aryan,&lt;br&gt;&lt;br&gt;thanks for the comments.  When you look at the specialized workloads that GPGPUs can provide additional processing power, it&#039;s amazing that they haven&#039;t been integrated sooner.  I downloaded an interesting whitepaper entitled &quot;QP: A Heterogeneous Multi-Accelerator Cluster&quot; done by the University of Illinois at Urbana-Champaign that really highlighted the strengths of using both foundational CPUs (OS and programmatic interface) as well as specialized GPGPUs for determined workloads.  I think, based on what you&#039;ve written above, that the encryption (data security) vertical is well suited to using these devices.  From a storage perspective, encrypt/decrypt, deduplication (calculation and stream intensive), and even more complex parity workloads will benefit highly from GPGPUs.&lt;br&gt;&lt;br&gt;Thanks again!&lt;br&gt;&lt;br&gt;cheers,&lt;br&gt;&lt;br&gt;Dave Graham</description>
		<content:encoded><![CDATA[<p>Aryan,</p>
<p>thanks for the comments.  When you look at the specialized workloads that GPGPUs can provide additional processing power, it&#39;s amazing that they haven&#39;t been integrated sooner.  I downloaded an interesting whitepaper entitled &#8220;QP: A Heterogeneous Multi-Accelerator Cluster&#8221; done by the University of Illinois at Urbana-Champaign that really highlighted the strengths of using both foundational CPUs (OS and programmatic interface) as well as specialized GPGPUs for determined workloads.  I think, based on what you&#39;ve written above, that the encryption (data security) vertical is well suited to using these devices.  From a storage perspective, encrypt/decrypt, deduplication (calculation and stream intensive), and even more complex parity workloads will benefit highly from GPGPUs.</p>
<p>Thanks again!</p>
<p>cheers,</p>
<p>Dave Graham</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Aryan</title>
		<link>http://flickerdown.com/2008/11/future-storage-systems-part-6-gpu-accelerated-storage/comment-page-1/#comment-41</link>
		<dc:creator>Aryan</dc:creator>
		<pubDate>Wed, 12 Nov 2008 11:28:13 +0000</pubDate>
		<guid isPermaLink="false">http://flickerdown.com/?p=259#comment-41</guid>
		<description>I expect GPGPU functions to be integrated into both software and hardware.&lt;br&gt;Software at first specialised apps later into the OS, hardware more and more into southbridges/hybrid solution. Ye ole M$ is barking at the tree in this regard, but progress in this field has been so storming that GPGPU integration is only a small problem compared with the huge security risks that are caused by this massive increase in avaliable processing power; a workload that a CPU took years is reduced a weeks. Ie: bruteforcing strong MD5 hashes. [ now listening GSYBE - Static ] Imagine a marriage between rainbow and bruteforce at this level. It takes me seconds to bruteforce any MD5 hash of 8 or less characters/numbers....  If m$ KEEPS ON SLACKING  they&#039;ll also miss this train as it streams into the cloud.</description>
		<content:encoded><![CDATA[<p>I expect GPGPU functions to be integrated into both software and hardware.<br />Software at first specialised apps later into the OS, hardware more and more into southbridges/hybrid solution. Ye ole M$ is barking at the tree in this regard, but progress in this field has been so storming that GPGPU integration is only a small problem compared with the huge security risks that are caused by this massive increase in avaliable processing power; a workload that a CPU took years is reduced a weeks. Ie: bruteforcing strong MD5 hashes. [ now listening GSYBE - Static ] Imagine a marriage between rainbow and bruteforce at this level. It takes me seconds to bruteforce any MD5 hash of 8 or less characters/numbers&#8230;.  If m$ KEEPS ON SLACKING  they&#39;ll also miss this train as it streams into the cloud.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
