Page 1 of 1

Pre-Postprocess visualization crash

Posted: Wed Jun 03, 2015 6:44 pm
by pepemxl
I have a mesh with 4 million of hexahedral elements, when I try to visualize the postprocess the program crash.

Another case is in the preprocess when I try to visualize the materials of 10 million of hexahedral elements the program crash, sometimes at the moment to load the mesh
the program crash.

I test the program with a large quantity of memory and less elements but the problem still.

How many elements support GiD?

Re: Pre-Postprocess visualization crash

Posted: Wed Jun 03, 2015 7:40 pm
by abelcs
There is not a limit in the number of elements to be treated by GiD. The only limit is the memory of the computer where GiD is installed.
This should be a bug, or maybe your model is somehow corrupted.

Could you please give us access to the model in order to try to reproduce this problem?

ABEL

Re: Pre-Postprocess visualization crash

Posted: Thu Jun 04, 2015 11:07 am
by escolano
To handle big result files you could enable in Utilities->Preferences "Use resulsts cache", and specify the allowed size for results (e.g. set a number like 1024 MB or 2048 MB)
Then results are loaded/unloaded on demand, when the user try to draw them.

Setting also 'Use indexed' an auxiliary index file will be created to a random access to the results file (then read the result definitions without load its data will be faster)


On a x32 platform the maximum about of RAM of a process is of 2 or 3 GB (the maximum memory direction is 2^32 =4GB, but about 1GB is reserved for the operating system).
For a bigger process a x64 platform must be used (with enouth physical RAM to store it)

Re: Pre-Postprocess visualization crash

Posted: Tue Jun 16, 2015 2:31 am
by pepemxl
abelcs wrote:There is not a limit in the number of elements to be treated by GiD. The only limit is the memory of the computer where GiD is installed.
This should be a bug, or maybe your model is somehow corrupted.

Could you please give us access to the model in order to try to reproduce this problem?

ABEL
Yes, but I only have a .msh file, I have generated models with less elements, and there is not problem, when I increase the quality the problem is presented.
escolano wrote:To handle big result files you could enable in Utilities->Preferences "Use resulsts cache", and specify the allowed size for results (e.g. set a number like 1024 MB or 2048 MB)
Then results are loaded/unloaded on demand, when the user try to draw them.

Setting also 'Use indexed' an auxiliary index file will be created to a random access to the results file (then read the result definitions without load its data will be faster)


On a x32 platform the maximum about of RAM of a process is of 2 or 3 GB (the maximum memory direction is 2^32 =4GB, but about 1GB is reserved for the operating system).
For a bigger process a x64 platform must be used (with enouth physical RAM to store it)
I have used both with x86 and x64, I watch the RAM memory and never reach the maximum. After enable in Utilities->Preferences "Use resulsts cache",
this fixed a part of the problem, I have loaded the mesh without problems after I have increased Utilities->Preferences "Use resulsts cache", and I can watch
the results very well but when I try to rotate the model in the porst process or draw the materials in the preprocess GiD Abandoned the visualization. I suppose the
problem in the postprocess can be fixed changing some parameters in the visualization, but in the draw of materials, I do not have idea.

Re: Pre-Postprocess visualization crash

Posted: Tue Jun 16, 2015 9:58 am
by escolano
In Utilities->Preferences... Graphical->System - Drawing method try to set 'Immediate mode' or 'Display list', this is slower drawing than "Vertex array" or "Vertex buffer objects", but require less memory.

If you are not using the last GiD developer version (12.1.6d) or last official (12.0.6), try to use it.

Which platform do you have? Windows, Linux, Mac OS X?
How much RAM is using the gid.exe process just before crashing?

If do you want, you can provide us a copy of your big mesh and results and we coult try to do some test to try to enhance the memory bottlenecks that appear in your case.