[FLASH-USERS] Performance numbers

[Sean Couch]

Hi Kevin,

This is pretty delayed so I hope it is still useful to you…

Performance is very machine- and problem-dependent, but for TACC Stampede, a more or less commodity Intel cluster, I have measured FLASH performance for the unsplit hydro solver using PARAMESH (;-)) AMR at about 1.0e-8 core-hours per zone per step in 3D.  And this weak scales nearly perfectly up to at least 4096 cores. YMMV.

Cheers,
Sean


-----------------------------------------------------------
Sean M. Couch
Sr. Postdoctoral Scholar
Theoretical Astrophysics Including Relativity
California Institute of Technology
Mail Code 350-17
Pasadena, CA  91125
(626) 395-4282
www.flash.uchicago.edu/~smc

On Oct 28, 2014, at 9:19 AM, Kevin Olson <Kevin.M.Olson at drexel.edu> wrote:

> Does anyone have any recent performance numbers for FLASH 4.2?
> 
> Thanks in advance,
> Kevin Olson

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20141101/255d003a/attachment.htm>