[Wien] Wien2k with load-distributing systems

Jorissen Kevin Kevin.Jorissen at ua.ac.be
Wed Sep 17 12:20:54 CEST 2003

I'd like to know what you think about the multithreading of the ifc+mkl
on shared memory systems.  They claim to have "excellent scaling",
whatever that may mean.


-----Oorspronkelijk bericht-----
Van: Peter Blaha [mailto:pblaha at zeus.theochem.tuwien.ac.at] 
Verzonden: woensdag 17 september 2003 11:26
Aan: wien at zeus.theochem.tuwien.ac.at

Finegrain mpi parallelization works ONLY reasonable when your matrixsize
is above 5000 (better 10000 or more!!!), and you need very fast
communication!! (shared memory), but even then you will never get 100%
speedup. If you have bad luck (slow communication, syncronization
problems), you might run on 4 processors NOT faster than on ONE !!

More information about the Wien mailing list