Hi, I just ran make_average_subjects with increased swap (from 2G to 8G) on 178 subjects and now it worked, which would seem to indicate that it may have been a memory issue? I did not crash with 3, 10 and 113 subjects with 2G swap.
yours, LMR
Hmmm, I've run it on >100 subjects without trouble. I'll try now and see if I can replicate locally. Is there anything different about these volumes? Are they all 256^3 and unsigned char?
cheers, Bruce
Could be a memory leak. Can you send us all the text output? We need to
know specifically what program it is dying on.
doug
Hi, I am analysing a dataset with 180 subjects. When I run
make_average_subjects it crashes, the error message usually >>>being something like this:
mghRead: encountered error executing: 'zcat /space/monkeys/1/home/lmr /subjects/VETSA_test/testAVG/tmp/make_average_vol-tmp/orig-19963b.mgz',frame
-1
I have tried running it with fewer subjects and it worked with 3, 10 and
73 subjects. Also, it crashes at different points in >>>the processing stream, so once it crashed at subject # 130, which I removed from the list of subjects, and then next time >>>it crashed at # 30. My frist guess was that it is a memory problem. I tried running the top command in a different
terminal at the same time as running make_average_subject with 2 and 10
subjects, and it seemed to require up to >>>400m of virtual memory and 10% of memory, but I couldn't see any difference in memory demands between running 2 >>>and 10 subjects. This was running with 4G of physical memory and 2G of swap.
Is this a known issue? (Couldn't find anything in the mailing list
archive..)
Thank you!