[Beowulf] Maximizing intra-node communication performance
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.
Jakob Oestergaard jakob at unthought.netTue Dec 27 17:56:50 PST 2005
- Previous message: [Beowulf] Maximizing intra-node communication performance
- Next message: [Beowulf] Maximizing intra-node communication performance
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Tue, Dec 27, 2005 at 12:05:55PM +0200, Tahir Malas wrote: > Hi all, > Taking advice from a previous discussion, we have purchased an Tyan server > with 8 dual-core Opteron 870 processors. Now I want to wonder how I can > maximize the intra-node communication of the server. We have been using > LAM-MPI, but I think that TCP/IP protocol may degrade the performance. Has When Linux does TCP on the local host only, that should be very very efficient (the kernel can make a lot of shortcuts and it does). Of course, there are more efficient ways, but I wanted to say that TCP on the local host is quite different performance wise that TCP over an actual external network. > anybody tried new implementations of MPI, or anybody knows some other > support for intra-node communication? Can't you tell MPI to use shared memory? (I'm sure that I used an MPI that allowed that - I think all I had to do was set an environment variable - but this is a long time ago and I don't remember the details exactly). -- / jakob
- Previous message: [Beowulf] Maximizing intra-node communication performance
- Next message: [Beowulf] Maximizing intra-node communication performance
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
