Version of gmake for clusters...
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.
lowther at att.net lowther at att.netThu Apr 12 04:08:40 PDT 2001
- Previous message: Version of gmake for clusters...
- Next message: Version of gmake for clusters...
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Andreas Boklund wrote: > > ----------------------------------------------------- > Number of time Did they fill the > computers minutes head's NIC > 1 5.58 no > 3 2.59 no > 5 3.30 yes > 7 3.10 yes > 9 2.30 yes > ----------------------------------------------------- > If anyone have any ideas on why the numbers are so irregular, > i would like to hear your theories. > It would appear as if your network was the bottle neck if by "Did they fill the head's NIC" means you were utilizing full bandwidth. You can add 1+1 9 times much faster on one processor than trying to push it out as sub processes to 9 machines. The processor overhead is much less than the network overhead. -- Ken Lowther Youngstown, Ohio http://www.atmsite.org
- Previous message: Version of gmake for clusters...
- Next message: Version of gmake for clusters...
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
