[Wien] hardware for wien2k

[Laurence Marks]

Comments:
1) It sounds reasonable. Get your vendor(s) to allow you to benchmark
on different systems. This will not just give you ideas about
comparable performance, it will tell you which ones you can trust.
2) You will get reasonably good performance for matrix sizes up to
about 18K if you run with 8 nodes per k-point. If you need to go much
larger (e.g. 24K up) you will need infiniband or equivalent.

On Tue, Jun 30, 2009 at 8:53 AM, Anne-Christine
Uldry<anne-christine.uldry at psi.ch> wrote:
> Dear Wien2k community,
>
> My group is considering buying some hardware that would be dedicated mostly
> to running Wien2k as efficiently as possible. We are working on iron-based
> magnetic systems. Typically, we would like to look at say 128 atoms
> supercells. Our computer officer suggested buying 5 blades with each 2
> quadcore 2.5GHz Intel E5540 CPUs and 24 GB of RAM (gigabit
> interconnet...infiniband might be added later, but I think I will stick to
> k-point parallelisation only). So that would give us 40 cores with 3GB of
> memory per core. Does this sound like a sensible choice for the type and
> size of system we want to look at ?
>
> Kind regards
> Anne-Christine Uldry
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>



-- 
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
Web: www.numis.northwestern.edu
Chair, Commission on Electron Crystallography of IUCR
www.numis.northwestern.edu/
Electron crystallography is the branch of science that uses electron
scattering to study the structure of matter.