Hi Stephen,
There are a couple things. It looks like every traceback ends with
KeyboardInterrupt. Not sure what that's about, but it's difficult to
tell what's going on. I've created a patch to separate out the
tracebacks from each processor, but keep in mind that this *could*
cause a deadlock if not all of them signal at the same place. (An
alternative would be to print each traceback to a file and then exit,
which I'll put into trunk, actually, if this sounds like a good plan
to y'all.)
http://paste.enzotools.org/show/67/
It should be useful in your case.
What I would try, to separate out the memory issues, would be manually
cleaning each grid after you do, say, 10 sphere:
for g in pf.h.grids: g.clear_data()
and see if that fixes it. If so, then it's a memory issue, and there
is likeyl a reference hanging around. On immediate inspection I
didn't see it, but I'll take another look. If you can get the
separated out tracebacks, that'd be awesome.
-Matt
On Mon, Mar 2, 2009 at 7:35 PM, Stephen Skory
Hi Matt,
I'm trying to run my code on L7, but it's crashing. It does a fair amount of the random spheres, and crashes like this:
http://stephenskory.com/err.txt
The version I'm using is here:
http://paste.enzotools.org/show/66/
I'm not asking you to fix this, Matt, but can you speculate on what might be happening? I am thinking it might be due to memory, which is why I am trying to 'del sp' in the script above, but that makes no difference. I printed out the last 100 or so centers that were chosen, and looked at them by hand to see if that was a problem, but none gave me any problems. I guess what I'm worried about is if what is causing this error will go away with your changes to this method? And therefore how much time should I spend trying to track it down?
Thanks!
_______________________________________________________ sskory@physics.ucsd.edu o__ Stephen Skory http://physics.ucsd.edu/~sskory/ _.>/ _Graduate Student ________________________________(_)_\(_)_______________ _______________________________________________ Yt-dev mailing list Yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org