<html><head><style type='text/css'>p { margin: 0; }</style></head><body><div style='font-family: Arial; font-size: 12pt; color: #000000'><P><BR>----- Original Message -----<BR>From: "Greg Lindahl" <lindahl@pbm.com><BR><BR>>Hey! In year 4 it's about the same to keep the old cluster, or throw<BR>>it out and buy a new one. (If I want to spend the same $$, I can buy a<BR>>cluster 1/4 the size, but same performance, as the original one.)<BR>><BR>>And in year 5, it's a big win to buy the new stuff.</P>
<P> </P>
<P>While Greg's analysis makes sense, one should probably note that</P>
<P>transistor-count-driven performance improvement assumptions yield</P>
<P>smaller practical benefits today where bandwidth continues to lag and</P>
<P>parallel performance improvement is not automatic. A gain in throughput</P>
<P>performance is guaranteed by the Moore's Law doubling in the multicore</P>
<P>era if your workflow is throughput oriented.</P>
<P> </P>
<P>;-)</P>
<P> </P>
<P>rbw<BR><BR><BR>_______________________________________________<BR>Beowulf mailing list, Beowulf@beowulf.org<BR>To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf<BR></P></div></body></html>