Racks vs. pile of PCs
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 Aug 13 20:54:49 PDT 2002
- Previous message: Racks vs. pile of PCs
- Next message: Racks vs. pile of PCs
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
> would you like to consider *blade server* for your evaluation? > http://www.nexcom.com.tw/ aren't those just RLX blades? but to answer your question: no, blades are not generally interesting to beowulf people, simply because they trade too much performance for density. PIII's just don't cut it against K7's and P4's for most (all?) compute intensive work. to be honest, I can't really think of an application that requires a separate CPU for each node. even a moderately heavy-duty machine can maintain "Enterprise" levels of web-hosting, for instance. maybe something that's extremely cache-friendly and not FP-intensive. but blades really are a heat and/or density sell. around 1/7 U and 10W per CPU, versus 1/2 U and 50-70W per CPU. might still be worth it, except that blades simply do not come close to $1K/CPU that duals (or small-form-factor unis) do. personally, I still like the idea of stacking small-for-factor unis on a wire rack. we're not taking that approach in our current cluster purchase, but that's mainly because I had an attack of conservativism ;)
- Previous message: Racks vs. pile of PCs
- Next message: Racks vs. pile of PCs
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
