Author Topic: pfeap_on not true under isw=1  (Read 11104 times)

d.brands

  • Jr. Member
  • **
  • Posts: 19
pfeap_on not true under isw=1
« on: March 03, 2022, 05:57:30 AM »
Dear all,

I figured out, that the global variable pfeap_on is not set to true under isw=1 during a parallel MPI run of ParFEAP.
Is this a "normal" behavior? If yes, which variable should used instead of pfeap_on under isw indicating a parallel execution?

Thanks in advance

Bests
Dominik

Prof. R.L. Taylor

  • Administrator
  • FEAP Guru
  • *****
  • Posts: 2649
Re: pfeap_on not true under isw=1
« Reply #1 on: March 03, 2022, 07:21:17 AM »
Can you explain why you need it on?  Isw=1 is for setting material set data

d.brands

  • Jr. Member
  • **
  • Posts: 19
Re: pfeap_on not true under isw=1
« Reply #2 on: March 03, 2022, 01:23:03 PM »
I want to open an additional output file under isw=1. In the parallel "mode" it should be open for each parallel instance.
In case of parallel "mode", the filename should be extend by the rank-id.

Prof. S. Govindjee

  • Administrator
  • FEAP Guru
  • *****
  • Posts: 1164
Re: pfeap_on not true under isw=1
« Reply #3 on: March 03, 2022, 02:39:48 PM »
The flag is not set until the PETSc macro command is issued.    So the default is a serial run.
If you want to force it earlier, I suggest creating a UMESh command.  That will be the simplest
non-intrusive way to accomplish this.  Then just place your umesh command very early in you input file.

Prof. S. Govindjee

  • Administrator
  • FEAP Guru
  • *****
  • Posts: 1164
Re: pfeap_on not true under isw=1
« Reply #4 on: March 03, 2022, 02:45:36 PM »
Another option is to program your element to check ntasks.gt.1  from setups.h in which case you at least know that mpirun has been issues and there are more than one task.

d.brands

  • Jr. Member
  • **
  • Posts: 19
Re: pfeap_on not true under isw=1
« Reply #5 on: March 03, 2022, 03:09:13 PM »
Thank you for the explanations.
Now, I have used the "ntasks.gt.1" way and it works