Pizza.py dump not reading all snapshots

Submitted by jwarnett on Tue, 05/10/2011 - 16:48

Hey

For some reason the dump command doesn't seem to want to read in all snapshots from my dump file

It reads in 21 snapshots, and ignores the others. There are infact 12000 snapshots! So why is it missing them off and how can I resolve this. Its quite a large simulation (dump file 2Gb!!), so is it struggling with the large file size?

Thanks

Jay

jwarnett | Thu, 05/12/2011 - 10:45

OK, ive noticed that I have 64bit CPUs and have installed a 32bit operating system... whoops. Probably a basic question... why would running the system at 64 bit correct this issue? Looks like I'll be reformatting here!

ckloss's picture

ckloss | Thu, 05/12/2011 - 10:52

afaik, depending on the linux kernel you are using, the per-process memory limit in a 32 bit system can be 2-4 GB
If the dump file is 2GB, standard pizza will need approx. 4 GB to post-process it b/c it opens the dump file and the files it writes to

Christoph

jwarnett | Thu, 05/12/2011 - 12:21

OK.

What is the best way to get LIGGGHTS to split the file as it is dumping.

Obviously I could use tmp.dump.* to get it to dump each snapshot, but this would mean 12000 files which would be far too many. Would be better if I could split 2000 dumps or so.

Is this as simple as

dump dmp all atom 1000 dump.rotatetankA
run 2000000
undump
dump dmp all atom 1000 dump.rotatetankB
run 2000000
etc...