It seems to me that there are only 2 ways for DX to screw up subsequent runs: to be using shared-memory segments and not freeing them, or by not actually going away and taking up virtual memory . Try running ipcs at the command line to see whats in use, and look in ps -elf for dxexec. I don't think it should be using shared memory, so I doubt thats the problem. Otherwise, I dunno. What are you trying to get DX to do? Any custom modules in there?
Greg
Michael Parker Owen
<[EMAIL PROTECTED]> To:
[email protected]
Sent by: cc:
[EMAIL PROTECTED] Subject: [opendx-general]
Memory leak under RHL
son.ibm.com
09/16/2002 04:05 PM
Please respond to
opendx-general
Hi,
I'm running OpenDX 4.2 under Red Hat Linux 7.3, and I seem to have a
nasty memory leak that chews up ~100 MB per network execution. I've read
some information on the -cache 0 option into the Image module, but this
doesn't seem to fix the problem (although it may help, the amount of
memory lost per exec seems smaller). Eventually DX runs out of memory,
and even quitting DX does not free all of this memory up (meaning that
each DX execution runs for a shorter and shorter time). After around 10
frames, I have to reboot.
Any thoughts, suggestions? This is an extremely critical roadblock, as I
cannot babysit the machine for the hundreds of cycles that are required
to render my images.
Thanks,
Michael Owen
Theoretical Astrophysics Group
North Carolina State University
Department of Physics
Box 8202
Raleigh, NC 27695-8202
