[Wien] Re: Two or one CPU's ?

Torsten Andersen thor at physik.uni-kl.de
Fri Feb 6 07:42:04 CET 2004


Whenever you have more than one k-point, k-point splitting is more 
effective since it minimizes the communication needed within the 
programs. Thus, it can also be used over networks (ethernet, myrinet, 
etc.) without any loss of efficiency. Memory bandwidth seems to be a big 
factor, so two 1-processor machines is likely faster than one 
2-processor machine - at least for Pentium/Xeon-based machines.

If you have just one k-point, then a multiprocessor machine is the 
fastest option (with MPI), since no network can compete with the 
communication speed of a crossbar switch.

Best regards,
Torsten Andersen

L. D. Marks wrote:
> As an extension of my prior email, has anyone investigated
> -parallel for 2 CPU machines, or is the k-point splitting
> more effective.
> 
> -----------------------------------------------
> Laurence Marks
> Department of Materials Science and Engineering
> MSE Rm 2036 Cook Hall
> 2225 N Campus Drive
> Northwestern University
> Evanston, IL 60201, USA
> Tel: (847) 491-3996 Fax: (847) 491-7820
> mailto:ldm at risc4.numis.northwestern.edu
> http://www.numis.northwestern.edu
> Nanocrystallography Workshop, http://ncem.lbl.gov/workshop.htm
> -----------------------------------------------
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> 

-- 
Dr. Torsten Andersen                     TA-web: http://deep.at/myspace/
AG Hübner, Department of Physics, Kaiserslautern University, and
Condensed Matter Theory Group, Department of Physics, Uppsala University
Web: http://www.fysik4.fysik.uu.se/         http://www.physik.uni-kl.de/ 






More information about the Wien mailing list