Skip navigation links
(Wind-US Documentation Home Page) (Wind-US User's Guide) (GMAN User's Guide) (MADCAP User's Guide) (CFPOST User's Guide) (Wind-US Utilities) (Common File User's Guide) (Wind-US Installation Guide) (Wind-US Developer's Reference) (Guidelines Documents)

(adf_revert) (cfaverage) (cfbeta) (cfcnvt) (cfcombine) (cflistnum) (cfpart) (cfreorder) (cfreset_iter) (cfspart) (cfsplit) (decompose) (fpro) (gpro) (jormak) (mpigetnzone) (recombine) ( (thplt) (tmptrn) (USintrpltQ) (windpar) ... (all utilities)


windpar may be used to compute an estimate of the potential for speed-up of a particular Wind-US case when run in parallel mode. The calculation is based on the likely number of grid points per processor, as determined from the number of zones and the grid size in each zone. Grid sizes are read from a common grid (.cgd) or common flow (.cfl) file. It is assumed that each processor is equally powerful, and zonal differences in solution procedure are not taken into account.


Suppose we have the simple three-zone configuration shown below, with grid sizes 17 × 6, 33 × 11, and 17 × 11 in zones 1, 2, and 3, respectively.

Three-zone grid

Input grid for windpar

Running windpar for this configuration gives the following results. Lines in slanted type are typed by the user.

   % windpar
                                ***** windpar *****
             Select the desired version from the following list.
   0) END
   1) windpar
   Single program automatically selected.
    windpar - Version 1.5 (last changed 2007/02/14 20:05:48)
    * Warning: This software contains technical data whose export is           *
    * restricted by the Arms Export Control Act (Title 22, U.S.C., Sec 2751,   *
    * et seq.) or Executive Order 12470. Violation of these export-control     *
    * laws is subject to severe criminal penalties. Dissemination of this      *
    * software is controlled under DoD Directive 5230.25 and AFI 61-204.       *
    Please input CFL or CGD filename
    Proc Speedup Proc Speedup Proc Speedup Proc Speedup
       1   1.000    2   1.796    3   1.796
    Maximum speedup attainable is
    a speedup of   1.796 on   2 processors
    with an efficiency of  89.807%
     Proc   Eff.  Proc   Eff.  Proc   Eff.  Proc   Eff.
       1 100.000    2  89.807    3  59.871

The speedup is defined as the total number of grid points in the problem, divided by the maximum number of grid points on any processor. The efficiency is then defined as the speedup divided by the number of processors, expressed as a percentage.

For this particular three-zone case, we have

Number of Processors   Points on Processor 1   Points on Processor 2   Points on Processor 3   Speedup   Efficiency
1 652 - - 1.000 100%
2 363 289 - 1.796 89.807%
3 363 187 102 1.796 59.871%

Thus, the best scenario for this case is to run on two processors. Because the total number of grid points in zones 1 and 3 combined (289) is less than the total number in zone 2 (363), the maximum speedup is the same whether two or three processors are used.

Last updated 7 Aug 2007