|
[Sponsors] |
March 26, 2017, 10:16 |
Writing animation in parallel
|
#1 |
Super Moderator
Alex
Join Date: Jun 2012
Location: Germany
Posts: 3,398
Rep Power: 46 |
For various reasons I have been using Paraview only in serial mode until now.
Today I thought I should give it another try. The goal: create an animation for a rather large transient simulation (~60 mio cells, 450 time steps). So I activated auto mpi, set the max number of cores to 22 (machine has 24 cores) and restarted paraview. Code:
AutoMPI: SUCCESS: command is: "/opt/paraview/ParaView-5.3.0-Qt5-OpenGL2-MPI-Linux-64bit/lib/paraview-5.3/mpiexec" "-np" "22" "/opt/paraview/ParaView-5.3.0-Qt5-OpenGL2-MPI-Linux-64bit/lib/paraview-5.3/pvserver" "--server-port=51753" AutoMPI: starting process server -------------- server output -------------- Waiting for client... AutoMPI: server successfully started. Unfortunately, that is about all I got. I can not switch to any other time step. Writing an animation creates the first image for the initial time step and nothing else. The 22 instances of "pvserver" are all running at 100% CPU load even after loading the initial time step is finished. Processing a single time step in serial mode takes around 11 minutes. I waited longer than that. Paraview version is the latest release 5.3.0, operating system is Opensuse Leap 42.1. Edit: same negative result with Paraview version 5.2.0. Any suggestions are welcome. Last edited by flotus1; March 27, 2017 at 05:25. |
|
March 28, 2017, 10:40 |
|
#2 |
Super Moderator
Alex
Join Date: Jun 2012
Location: Germany
Posts: 3,398
Rep Power: 46 |
A small bump with some new information.
Tweaking the ensight gold case file to include file set and time set for the geometry, I at least managed to get an error message out of paraview. It loads the first time step, but when I try to switch to a different time step I get this: pv_error.png What I am very suspicious about is the "0" added after the geometry path. Does ParaView expect some kind of parallel file format? Edit: just to clarify, doing the same steps in serial mode works like a charm with no error messages. So my data structure can not be entirely wrong. |
|
April 27, 2017, 11:55 |
|
#3 |
Super Moderator
Alex
Join Date: Jun 2012
Location: Germany
Posts: 3,398
Rep Power: 46 |
Problem solved.
My case files looked something like this: Code:
FORMAT type: ensight gold GEOMETRY model: ../ModelEnsight/LB_results.geo VARIABLE scalar per element: 1 1 Pressure LB_results_primary_Scalar.scl vector per element: 1 1 Velocity LB_results_primary_Vector.vec TIME time set: 1 number of steps: 3 time values: 0 1 2 FILE file set: 1 number of steps: 3 Code:
FORMAT type: ensight gold GEOMETRY model: 0 ../ModelEnsight/LB_results.geo VARIABLE scalar per element: 1 1 Pressure LB_results_primary_Scalar.scl vector per element: 1 1 Velocity LB_results_primary_Vector.vec TIME time set: 1 number of steps: 3 time values: 0 1 2 FILE file set: 1 number of steps: 3 |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Writing parallel non-partitioned ordered data (TecIO). | artu72 | Tecplot | 15 | March 13, 2019 08:34 |
[mesh manipulation] Dynamic remeshing (mequite) in parallel not working [foam-extend-4.0] | Peter_600 | OpenFOAM Meshing & Mesh Conversion | 4 | August 1, 2017 06:07 |
Running parallel case after parallel meshing with snappyHexMesh? | Adam Persson | OpenFOAM Running, Solving & CFD | 0 | August 31, 2015 22:04 |
Script to Run Parallel Jobs in Rocks Cluster | asaha | OpenFOAM Running, Solving & CFD | 12 | July 4, 2012 22:51 |
Writing parallel code | hartinger | OpenFOAM Running, Solving & CFD | 4 | March 13, 2006 13:07 |