[Beowulf] Network considerations for new generation cheap beowulfcluster
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.
Larry Stewart larry.stewart at sicortex.comTue May 22 06:02:38 PDT 2007
- Previous message: [Beowulf] Network considerations for new generation cheap beowulfcluster
- Next message: [Beowulf] Network considerations for new generation cheap beowulfcluster
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Eugen Leitl wrote: >On Tue, May 22, 2007 at 12:38:13AM -0400, Mark Hahn wrote: > > > >>unless you're suggesting that games use collectives akin to MPI >>broadcast/reduce/etc? >> >> > >I could see how a low-latency interconnect would help with a >Second Life-like virtual reality, where most things happen >server-side. On the other hand, given that low-end servers >come with 4 GBit Ethernet NICs, just distributing the game >world across a 2d torus would do as well, or better. > >Has anyone ever heard of a cluster wired in a diamond >lattice (made from tetrahedrons) topology? > > > I haven't heard of anyone using a diamond lattice to wire a cluster. We're using a Kautz graph, which already makes my brain hurt. What would the advantages of a diamond lattice be? In terms of bisection and diameter? Ease of wiring? -Larry -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.scyld.com/pipermail/beowulf/attachments/20070522/f7631652/attachment.html
- Previous message: [Beowulf] Network considerations for new generation cheap beowulfcluster
- Next message: [Beowulf] Network considerations for new generation cheap beowulfcluster
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
