Message: Help desperately needed on the tessellated GDML volumes. Not Logged In (login)
 Next-in-Thread Next-in-Thread
 Next-in-Forum Next-in-Forum

None Help desperately needed on the tessellated GDML volumes. 

Forum: Geometry
Date: 27 Apr, 2012
From: Yipeng Liu <Yipeng Liu>

Hello all!

Recently I encountered a serious problem which has bothered me for a month and basically paralyzed my working progress on geometry implementation. The thing is, I used Fastrad and CAD .step files for our new detector to generate about 137 tessellated GDML volumes (files/modules). And I tried to include them into G4. It worked with part of those volumes, but with all of them or most of them the program (just to draw geometry, no particles) crashed by throwing the "bad_alloc" message indicating there is not enough (virtual) memory for my job on our computer cluster. And I used 2 parts of the exported GDML volumes which each has perfectly worked without crash, I read them at the same time into G4 and ran the macro but the crash occurred! So that excludes the possibility of a "bug" in a GDML file which can cause this crash.

So I thought the tessellated geometry is just simply too large (compared to the ones made by G4 C++ codes or manually written with GDML), every volume like a screw contains so many vertices and facets and if I didn't get it wrong, the GDML reader just creates the "in-memory" geometry from the parsed in GDML files. So all those "complex" geometrical info is just stored in the memory? So I just tried our most powerful computing cluster in our local research center which has 6 GB virtual ram for each job and it failed again! I also attached the screenshot of the error message which is always the same.

And I don't think it has anything to do with the stack volume either. 'cuz the stack limit for my account was largely increased but nothing was improved. I tried just one more GDML volume and it still crashed.

So I wonder if other G4 users who have experiences with Fastrad produced GDML volumes or any other software generated tessellated GDML volumes have had such problems with (virtual) memory. I guess our largest computer cluster is already quite competent and I don't know what else I can do to overcome this problem. By the way someone suggested there might be a memory leak. But what and where is that exactly? Is it really the cause or??

   Attachment:
      http://hypernews.slac.stanford.edu/HyperNews/geant4/get/AUX/2012/04/27/03.03-37593-ror_on_all.q_of_Tesla.png

Inline Depth:
 1 1
 All All
Outline Depth:
 1 1
 2 2
 All All
Add message: (add)

1 None: Re: Help desperately needed on the tessellated GDML volumes.   (Gong Ding - 11 May, 2012)
 Add Message Add Message
to: "Help desperately needed on the tessellated GDML volumes."

 Subscribe Subscribe

This site runs SLAC HyperNews version 1.11-slac-98, derived from the original HyperNews


[ Geant 4 Home | Geant 4 HyperNews | Search | Request New Forum | Feedback ]