VECFEM3 Reference Manual: vemrun

Type: command


vemrun - runs the executable a.out generated by VECFEM


vemrun [<nproc> [<executable>]]


vemrun starts the executable <executable> on a scalar computer as well as on a parallel computer with <nproc> processors. If <executable> is not specified a.out is assumed. The input file of the mesh data have to be in the current directory. The output files of the solution and the error estimator are written to the same directory, where the mesh data are read. If <nproc> is omitted one processor is used. If the executable program is specified, <nproc> has to be specified, too. The messages printed by the VECFEM routines can be analysed by the vemhint tool. vemrun has to be called from the ksh shell.


The following statement starts a.out on one processor:

    ~> vemrun

The following statement starts a.out on six processor:

    ~> vemrun 6

The following statement starts the executable test in the home directory:

    ~> vemrun 6 ~/test


The shell variables $VECFEM_ARCH and $VEMRUN_OPTIONS are set by the shell script vempfade, which is started by vemrun (see vemcompile). $VEMRUN_OPTIONS specifies additional options for running the executable (e.g. buffer sizes).


The input and output files are read and written from the solution sets and error estimator are used in the current directory but only on processor 1. Additionally three temporary swap files in /tmp/* are locally used on each processor.


  • IBM: On the IBM SP ($VECFEM_ARCH=AIXSP) the executable is started in the poe environment. To specify options of poe (eg. the used pool, network, protocol) use the correspending shell variables of the poe. Don't forget to export them!


VECFEM, xvem, vemcompile, vemhint, vembuild, vemexamples


Program by L. Grosz, 1995-1996. Copyrights by Universitaet Karlsruhe 1989-1996. Copyrights by Lutz Grosz 1996. All rights reserved. More details see VECFEM.