CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Bugs (
-   -   ParaView fills my RAM (

Phicau December 7, 2012 04:57

ParaView fills my RAM
Hi all,

it is well known that ParaView fills up your RAM little by little while you are exporting a video (even snapshot by snapshot), but this other thing I'm reporting is new for me...

I am running ParaView on a cluster, exporting the X server, so no RAM should be being used in my computer. However while I am exporting a video my RAM is slowly being used. The process Xorg is taking it. In the end my computer (8GB) starts to swap and becomes very slow.

It does not matter if ParaView window is on the screen, if it is hidden behind other windows or if it is minimized, I end up having nearly 30 snapshots exported and my computer needing to be restarted. ParaView is only taking 16GB on the cluster, which has 64GB available.

A little technical info:

Cluster running RHL 6, OpenFOAM 2.1.1, ParaView 3.10.0. OF & PV 64 bits, installation using CentFOAM project.

Local computer: Running Ubuntu 10.04, 64 bits, fully updated.

Did anyone encounter the same problem? Any tentative solutions?


startingWithCFD December 7, 2012 07:32

Same thing happens with me.

SUSE Linux Enterprise Server 11 SP1 (x86_64)
OpenFOAM v2.1.0
ParaView v3.12.0
Installation from sources.

Local computer
OpenSUSE 11.4 32 bits
Fully updated (I think).

EDIT: I usually end up with more snapshots, it depends on the grid size clearly.
But since I am interested in plotting only a specific patch of my mesh, I load only that in memory and work around the limitation.

kmooney December 7, 2012 14:55

Download the newest release candidate from Kitware (3.98.0). I no longer had this issue once I updated. I used the out of the box binary as well.

Phicau December 10, 2012 04:18

No issues for now with this version, thanks a lot!

However, it's 10 times slower, so I guess I will end up having to compile this version for my system.

kmooney December 10, 2012 12:04

That's too bad about the performance hit. I was actually surprised that the vanilla binary worked as fast as it does. Perhaps I'll try to do my own compilation and see if it gets better.

All times are GMT -4. The time now is 22:37.