Myrinet scalability
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.
Ole W. Saastad ole at scali.comSun Jun 16 01:53:45 PDT 2002
- Previous message: PORTLAND COMPILERS
- Next message: Setting up nodes with built in hard disk
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hi, with this talk about scalability and switches I would like to point out that the SCI interconnect uses no switch. The only thing you need to add an extra compute nodes and just recable the cluster. The cost increases linearly with the number of nodes. There are no step costs when you must buy more switch ports. Each of the SCI cards are in fact a switch and the switch in then distributes over the whole cluster and thence scale with the number of nodes (Scalable Coherent Interface). In this way you will always have full bisection bandwidth and extremely low latency throughout the cluster. Try running the benchmark all2all on different clusters and interconnects. Technical information : http://www.scali.com/Technology/index.shtml http://www.dolphinics.com/ -- Ole W. Saastad, Dr.Scient. Scali AS P.O.Box 150 Oppsal 0619 Oslo NORWAY Tel:+47 22 62 89 68(dir) mailto:ole at scali.no http://www.scali.com Are you meeting Petaflop requirements with Gigaflops performance ? - Scali Terarack bringing Teraflops to the masses.
- Previous message: PORTLAND COMPILERS
- Next message: Setting up nodes with built in hard disk
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
