<?xml version="1.0" encoding="utf-8" ?>
<rss version="2.0" xmlns:osnews="http://www.osnews.com/rss2#">
	<channel>
		<title>OSNews: </title>
		<link>http://www.osnews.com/story/13977/Sun_s_Rock_Goes_16_Cores</link>
		<description>Exploring the Future of Computing</description>
		<language>en-us</language>
		<copyright>Copyright 2001-2013, David Adams</copyright>
		<webMaster>adam+nospam@osnews.com</webMaster>
		<lastBuildDate>Sun, 19 May 2013 22:53:59 GMT</lastBuildDate>
		<image>
			<url>http://www.osnews.com/images/osnews.gif</url>
			<title>OSNews.com</title>
			<link>http://www.osnews.com</link>
		</image>
		<item>
			<title>T1 also has 32 threads</title>
			<link>http://www.osnews.com/thread?104475</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104475</guid>
			<description>8 x 4, that is.<br />
<br />
But of course, Rock is a much more powerful CPU. However, it's a bit worryingly over-the-horizon.</description>
			<pubDate>Wed, 15 Mar 2006 00:13:00 GMT</pubDate>
			<author>donotreply@osnews.com (mario)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE: T1 also has 32 threads</title>
			<link>http://www.osnews.com/thread?104487</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104487</guid>
			<description>Not really.  The first-generation Opteron and Conroe quad-cores will be out in 2007, and the Itanium Montecito in the same time frame as well.  Rock will come out just a little bit later, and will have the advantages of - A) Being supported by a very SMP/Multi-threaded operating system and B) Being based off of the by-then ultra-stable T1 architecture.<br />
<br />
Seriously, although I was hoping for a 2007 launch as well, this is still well within the window of being effective.</description>
			<pubDate>Wed, 15 Mar 2006 01:20:00 GMT</pubDate>
			<author>donotreply@osnews.com (Dark Leth)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>Jupiter</title>
			<link>http://www.osnews.com/thread?104488</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104488</guid>
			<description>I love The Reg's blatant self-promotion, especially since (too bad) they can't read their own stories before posting incorrect information.  The article talks about &quot;Jupiter&quot; as being a Sun UltraSPARC-T1 variant.  Yet, the article links to a similar story (and public knowledge) about how Jupiter is a Fujitsu chip, a follow on to their Olympus chip, which is what powers the upcoming APL line.<br />
<br />
What is this that the article says it has been since 2000 that Sun has &quot;garnered any accolades&quot;?  The US-IV/US-IV+ were pretty highly rated, and the T1 has been gathering accolades left and right.  I guess if you discount last year, then yep, it was 2000.<br />
<br />
Sun's follow on to the T1 is the NiagraII which will be 8 cores, 8 threads per core, with an FPU per core (instead of the current single FPU for the entire chip).</description>
			<pubDate>Wed, 15 Mar 2006 01:20:00 GMT</pubDate>
			<author>donotreply@osnews.com (spotter)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>Big bets</title>
			<link>http://www.osnews.com/thread?104546</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104546</guid>
			<description>Sun has really set themselves up to be the expert on high-thread-count-in-small-packages computing.  For a while I thought Rock was their new single-thread CPU, but I guess they are doing a different approach to Niagara-like CPUs.  Given Solaris' strong reputation in threading, sysadmins ought to like it.</description>
			<pubDate>Wed, 15 Mar 2006 05:53:00 GMT</pubDate>
			<author>donotreply@osnews.com (SeeMyNuts)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?104650</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104650</guid>
			<description>What really confuses me about the T1 is to figure out what runs good / what runs bad.<br />
<br />
The only thing I know is that it's supposedly brilliant for webhosting purposes.<br />
<br />
Say I wanna crunch data (with SQL? or similar), will it manage this good/great or bad? How do I know when the &quot;only one floating point CPU&quot; will damage the performance?</description>
			<pubDate>Wed, 15 Mar 2006 15:28:00 GMT</pubDate>
			<author>donotreply@osnews.com (Haicube)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE: Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?104662</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104662</guid>
			<description><a href="http://www.sun.com/blueprints/1205/819-5144.pdf" rel="nofollow">http://www.sun.com/blueprints/1205/819-5144.pdf</a><br />
<a href="http://blogs.sun.com/roller/page/rmc?entry=welcome_to_the_cmt_era" rel="nofollow">http://blogs.sun.com/roller/page/rmc?entry=welcome_to_the_cmt_era</a> <br />
<br />
Any use to you?</description>
			<pubDate>Wed, 15 Mar 2006 16:17:00 GMT</pubDate>
			<author>donotreply@osnews.com (JonAnderson)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE: Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?104665</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104665</guid>
			<description>My understanding is that the T1 is basically ideal for &quot;business&quot; workloads, such as web servers, database, file servers, etc.  It would be relatively bad as a graphics workstation or running analysis that uses a lot of floating point calculations.  So:  engineers and scientists probably want Opteron, and network and website administrators probably want T1.</description>
			<pubDate>Wed, 15 Mar 2006 16:20:00 GMT</pubDate>
			<author>donotreply@osnews.com (SeeMyNuts)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE: Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?104678</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104678</guid>
			<description>In a quick test using MySQL 5.0.18 and sysbench benchmark tool on both a T2000 (evaluation machine) and a V880, the T2000 easily beat the V880. If this is an example of the performance of the Niagara CPU, I can't wait for Rock!</description>
			<pubDate>Wed, 15 Mar 2006 17:02:00 GMT</pubDate>
			<author>donotreply@osnews.com (Robert Escue)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE[2]: Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?104774</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104774</guid>
			<description>If it doesn't use many fp-ops I don't think the difference would be that great.</description>
			<pubDate>Wed, 15 Mar 2006 20:42:00 GMT</pubDate>
			<author>donotreply@osnews.com (CaptainPinko)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE[2]: Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?104965</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104965</guid>
			<description>Okey what I would like to run at maximum performance is Enterprise Guide from SAS. Check this URL<br />
<br />
<a href="http://www.sas.com/technologies/bi/query_reporting/guide/" rel="nofollow">http://www.sas.com/technologies/bi/query_reporting/guide/</a> <br />
(Statistical software, handles a lot of multivariate stuff)<br />
<br />
And when we're talking floating point, I assume that means it has trouble with decimal figures, would that be a correct assumption? So for that purpose I'm better off with a AMD64?</description>
			<pubDate>Thu, 16 Mar 2006 07:40:00 GMT</pubDate>
			<author>donotreply@osnews.com (Haicube)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE[3]: Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?104997</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?104997</guid>
			<description>OK, you could 'Try before you buy'. If you already have<br />
this software running under Solaris or Linux you could<br />
try coolst from this page:<br />
<br />
<a href="http://cooltools.sunsource.net/nonav/index.html" rel="nofollow">http://cooltools.sunsource.net/nonav/index.html</a><br />
<br />
It does some heuristics on the running codes to see if<br />
they are suitable for deployment on T1.<br />
<br />
As an aside, if you are interested in Sun and Solaris<br />
it's worth periodically checkling the opensolaris pages.<br />
It's not just the code ;-)</description>
			<pubDate>Thu, 16 Mar 2006 10:22:00 GMT</pubDate>
			<author>donotreply@osnews.com (JonAnderson)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>Re: SAS-EG</title>
			<link>http://www.osnews.com/thread?105071</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?105071</guid>
			<description>Well, since the link says that SAS-EG is a &quot;Microsoft Windows client application&quot;, you wouldn't be able to run that on the T1 processor anyway.  The T1 is a SPARC processor, not an X86 processor.<br />
<br />
If you are talking about SAS products in general, they tend to be very heavy floating point based (statistical analysis), and would perform pretty poorly on the T1.</description>
			<pubDate>Thu, 16 Mar 2006 17:01:00 GMT</pubDate>
			<author>donotreply@osnews.com (spotter)</author>
			<category>Comments</category>
		</item>

		<item>
			<title>RE[3]: Maybe I'm tarded but...</title>
			<link>http://www.osnews.com/thread?105850</link>
			<guid isPermaLink="true">http://www.osnews.com/thread?105850</guid>
			<description>I don't know anything about SAS, but statistical analysis could be either integer math or floating point math, depending on what's being studied.  But, basically, floating point is math with decimal points and fractions, and integer math truncates off the fractions and works with whole numbers.  It turns out that most business/networking software uses integer math, which is what T1 is best at.</description>
			<pubDate>Mon, 20 Mar 2006 03:21:00 GMT</pubDate>
			<author>donotreply@osnews.com (SeeMyNuts)</author>
			<category>Comments</category>
		</item>
	</channel>
</rss>
