Subject: /. The Space Simulator Cluster
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Mark Hahn hahn at physics.mcmaster.caTue Nov 26 06:09:48 PST 2002
- Previous message: Subject: /. The Space Simulator Cluster
- Next message: Subject: /. The Space Simulator Cluster
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
> lanl's web page says the per node price was under $1k. So with 294 nodes and > a $500k total price, they've still got over $100k to spend on switches and > wire racks. the racks, of course, are dead cheap (costco, $100 or so). > effective to me. A 1U or 2U rack case is just as dense as those shuttle > boxes. hmm, 1U is 19" x 1.75 = 214.5 sq cm of "rack surface area". xpc is 20x28.5cm=370 sq cm an "ultraslim" desktop from HP (eg) is 218 sq cm. so yes, xpc's use somewhat larger space. that's hardly a bad thing, since if you were really interested in density, you'd be using blades. airflow might well be easier to manage with these. the real question is: how reliable are they? if you figure that they have two 50Khour fans in each unit (PS and CPU/heatpipe), they'll be replacing a unit every 3.5 days. no doubt they'll have a pile of spares, but it's still something of a pain. > With a dual CPU board you could get by with half the number of Gb > switch ports, which aren't cheap for >24 port switches. true, but your linpack score would probably also go down...
- Previous message: Subject: /. The Space Simulator Cluster
- Next message: Subject: /. The Space Simulator Cluster
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
