<?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: AMD: Teraflop (sic) in a box</title>
	<atom:link href="http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/feed/" rel="self" type="application/rss+xml" />
	<link>http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/</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: karenm</title>
		<link>http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/#comment-53</link>
		<dc:creator>karenm</dc:creator>
		<pubDate>Sat, 03 Mar 2007 04:15:55 +0000</pubDate>
		<guid isPermaLink="false">http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/#comment-53</guid>
		<description>So should we be saying Gigabits Ethernet?</description>
		<content:encoded><![CDATA[<p>So should we be saying Gigabits Ethernet?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Jerry</title>
		<link>http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/#comment-50</link>
		<dc:creator>Jerry</dc:creator>
		<pubDate>Fri, 02 Mar 2007 02:47:30 +0000</pubDate>
		<guid isPermaLink="false">http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/#comment-50</guid>
		<description>Your distinction between teraflop and teraflops, for the given context, is 100% accurate (the &quot;s&quot; is for seconds, *not* to pluralize flop).  However, we can say 1 teraflops, or we can say 1 teraflop/s, where teraflop means billion floating-point operations (an entirely different usage than AMD&#039;s).  Take a look at www.erdc.hpc.mil/systemNews/SGI_O3K/hardware, where the performance of their SGI is given in TFLOPS.  Now look at www.erdc.hpc.mil/systemNews/Cray_XT3/hardware, where the performance of their Cray is given in TFLOP/s.  Both usages are totally correct.  It is in this latter context that the term &quot;teraflop&quot; exists.  Note that the terms TFLOPS and TFLOP/s are dimensionally [floating-point operations/second], whereas TFLOP is dimensionally [floating-point operations].  Thus, *ALL* computers that contain floating-point hardware are &quot;Teraflop in a box.&quot;  Of course some systems will do that teraflop faster than others!

The basic issue is ignorance, which one would think could be cured via education.  In Mississippi, where I live, some folks think license is plural.  Example:  &quot;I ain&#039;t got my license; the judge took *them* away from me.&quot;  As it turns out, stupidity, that is, ignorance coupled with an unwillingness to be educated, cannot be cured.

A few years ago at work, a document was published wherein MB was defined as &quot;millions of bytes.&quot;  Thinking I was doing everyone a great favor, and being eager to establish my credibility (since I was the new kid on the block); I painstakingly constructed a document which carefully described why the definition was wrong.  I explained that the prefix, M, did sometimes mean 10^6, as in MHz, but that when used with bytes it meant, 2^20, as in 20MB.  I used several examples and sprinkled my diatribe with just enough satire to motive the folks who messed things up into fixing them.  The net result of all this is that I was accused of nit-picking (that&#039;s the nice version).  The final rationalization for their continuing stupidity was two-fold; 1) &quot;disk drive sales folks use millions of bytes so we should too,&quot; and 2) &quot;it&#039;s close enough.&quot;  It was quite aggravating.  Let me paraphrase the first justification, &quot;Billy is being stupid, so I&#039;ll be stupid too.&quot;  As to the second justification, I tried using that at the bank when they charged my wife $30 to cover her 13-cent overdraft (it *did not* work - the banking world has this silly insistence on accuracy).  As stated earlier, stupidity cannot be cured.

The moral of all this is that your engineering instinct and desire for accuracy in describing units of measure (in ALL venues) are not likely to make much of an impact on the farmers, politicians, sales folk, courtiers, and businessmen who make decisions about what gets published; essentially, you are a wart.  I do laud your courage for trying.  Please forward me a copy of the email that AMD sends when they acknowledge their ignorance and thank you for the education.  Neither of us should hold our breath while waiting.</description>
		<content:encoded><![CDATA[<p>Your distinction between teraflop and teraflops, for the given context, is 100% accurate (the &#8220;s&#8221; is for seconds, *not* to pluralize flop).  However, we can say 1 teraflops, or we can say 1 teraflop/s, where teraflop means billion floating-point operations (an entirely different usage than AMD&#8217;s).  Take a look at <a href="http://www.erdc.hpc.mil/systemNews/SGI_O3K/hardware" rel="nofollow">http://www.erdc.hpc.mil/systemNews/SGI_O3K/hardware</a>, where the performance of their SGI is given in TFLOPS.  Now look at <a href="http://www.erdc.hpc.mil/systemNews/Cray_XT3/hardware" rel="nofollow">http://www.erdc.hpc.mil/systemNews/Cray_XT3/hardware</a>, where the performance of their Cray is given in TFLOP/s.  Both usages are totally correct.  It is in this latter context that the term &#8220;teraflop&#8221; exists.  Note that the terms TFLOPS and TFLOP/s are dimensionally [floating-point operations/second], whereas TFLOP is dimensionally [floating-point operations].  Thus, *ALL* computers that contain floating-point hardware are &#8220;Teraflop in a box.&#8221;  Of course some systems will do that teraflop faster than others!</p>
<p>The basic issue is ignorance, which one would think could be cured via education.  In Mississippi, where I live, some folks think license is plural.  Example:  &#8220;I ain&#8217;t got my license; the judge took *them* away from me.&#8221;  As it turns out, stupidity, that is, ignorance coupled with an unwillingness to be educated, cannot be cured.</p>
<p>A few years ago at work, a document was published wherein MB was defined as &#8220;millions of bytes.&#8221;  Thinking I was doing everyone a great favor, and being eager to establish my credibility (since I was the new kid on the block); I painstakingly constructed a document which carefully described why the definition was wrong.  I explained that the prefix, M, did sometimes mean 10^6, as in MHz, but that when used with bytes it meant, 2^20, as in 20MB.  I used several examples and sprinkled my diatribe with just enough satire to motive the folks who messed things up into fixing them.  The net result of all this is that I was accused of nit-picking (that&#8217;s the nice version).  The final rationalization for their continuing stupidity was two-fold; 1) &#8220;disk drive sales folks use millions of bytes so we should too,&#8221; and 2) &#8220;it&#8217;s close enough.&#8221;  It was quite aggravating.  Let me paraphrase the first justification, &#8220;Billy is being stupid, so I&#8217;ll be stupid too.&#8221;  As to the second justification, I tried using that at the bank when they charged my wife $30 to cover her 13-cent overdraft (it *did not* work &#8211; the banking world has this silly insistence on accuracy).  As stated earlier, stupidity cannot be cured.</p>
<p>The moral of all this is that your engineering instinct and desire for accuracy in describing units of measure (in ALL venues) are not likely to make much of an impact on the farmers, politicians, sales folk, courtiers, and businessmen who make decisions about what gets published; essentially, you are a wart.  I do laud your courage for trying.  Please forward me a copy of the email that AMD sends when they acknowledge their ignorance and thank you for the education.  Neither of us should hold our breath while waiting.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: CC</title>
		<link>http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/#comment-49</link>
		<dc:creator>CC</dc:creator>
		<pubDate>Thu, 01 Mar 2007 22:17:16 +0000</pubDate>
		<guid isPermaLink="false">http://insidehpc.com/2007/03/01/amd-tflops-in-a-box/#comment-49</guid>
		<description>Violence inherent in the system, eh?</description>
		<content:encoded><![CDATA[<p>Violence inherent in the system, eh?</p>
]]></content:encoded>
	</item>
</channel>
</rss>
