testing openflower

Help
chemflux
2005-01-06
2013-04-09
  • chemflux
    chemflux
    2005-01-06

    I tested the performance of openflower using prisms_hexas.flw within examples folder
    I converted prisms_hexas.geo into *.msh using gmsh and excute OpenFlower command to combind with *.flw and *.msh
    Ahh!....
    I saw that the caculation results were going into out file..
    It takes too much time to complete procesure
    It was about 42 minute under Pentium 4 processor..
    Maybe we need to make cluster in order to perform our task well  
    Great!!
    The result of temperature profile was pretty good.
    It was my first CFD caculation!

     
    • Hi,
      It is good to hear that you succeeded the different steps in using OpenFlower ;-)

      Well, the prisms_hexas example file may not be optimized, I mean that the final time I gave in the input file maybe very large in terms of diffusion time...
      I hope that it's the case ?!!

      Anyway, I know that a little improvement must be done in the code since it is at its early stages. I already made a couple of months ago a (huge ! ) optimization saving up to (if I recall exactly) 25 % CPU time in computing diffusion... but I guess we could optimize it a little bit more.

      See you soon,
      VortexFlow.

       
    • chemflux
      chemflux
      2005-01-07

      Hello, VortexFlow
      your answer about calculation time makes sense.
      optimization process is time consuming and quite difficult processs.
      I hope that the performance will be improved through modifying optimation process.
      Thank you