OK, I might be misreading the memory statistics using the top command. I don't seem to be having the same problem with running out of memory as was the case earlier, so my problem might have been fixed with the recent changes to Brat. ...Steve > From: "Stephen J. Sanders" <ssanders@ukans.edu> > Reply-To: brahms-dev-l@bnl.gov > Date: Sun, 15 Oct 2000 09:21:18 -0500 > To: <brahms-dev-l@bnl.gov> > Subject: memory leak associated with running TestGeantDig > > Hi, I'm hoping one of the experts will have a suggestion on this: > I'm trying to use the TestGeantDig program to make the .cdat to > .root conversion in looking at some GBRAHMS results. There seems to be a bad > memory leak associated with running this program, however, and this is > preventing me from looking at some longer geant runs (I end up with an > out-of-memory segfault). To check that the problem is not associated > with one of my local source modifications, I just did a complete cvs > checkout of brat onto a rcas machine and ran the TestGeantDig routine on > a 500 event .cdat file with: > TestGeantDig -g -f -b -m /brahms/data01/simu/gbrahms_output/sim_562 > Using Top to monitor memory usage, I started out with 212,048K free. > After the program exited, I was down to less than 100,000K free. (Will > a reboot now be required to make this memory available again?) > > I've started to look for the problem and think I've traced things to > BrGeantInput, but I may be fooling myself. Since I need the geant > simulation results, I'll continue to look at this. I'm hoping, however, > that someone might be aware of the problem and know of a workaround or, > possibly, be able to suggest a likely area of the program where I should > be looking for the leak. ...steve >
This archive was generated by hypermail 2b29 : Sun Oct 15 2000 - 17:52:27 EDT