[Beowulf] RE: Capitalization Rates - How often should you replace a cluster? (resent - 1st sending wasn't posted ).

[richard.walsh at comcast.net]

----- Original Message ----- 
From: "Greg Lindahl" <lindahl at pbm.com> 

>Hey! In year 4 it's about the same to keep the old cluster, or throw 
>it out and buy a new one. (If I want to spend the same $$, I can buy a 
>cluster 1/4 the size, but same performance, as the original one.) 
> 
>And in year 5, it's a big win to buy the new stuff. 



While Greg's analysis makes sense, one should probably note that 

transistor-count-driven performance improvement assumptions yield 

smaller practical benefits today where bandwidth continues to lag and 

parallel performance improvement is not automatic.  A gain in throughput 

performance is guaranteed by the Moore's Law doubling in the multicore 

era if your  workflow is throughput oriented. 



;-) 



rbw 


_______________________________________________ 
Beowulf mailing list, Beowulf at beowulf.org 
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.scyld.com/pipermail/beowulf/attachments/20090121/3fc567f7/attachment.html