--cpulimit flag

36 views
Skip to first unread message

David Murphy

unread,
May 30, 2013, 3:43:51 PM5/30/13
to astro...@googlegroups.com
Dear all,

I am running a batch of fits files via "solve-field". There a fraction for which no solution is forthcoming, despite a long run-time. Looking closely at the flags, there is a "--cpulimit" (or "-l") option. The description suggests that after a defined number of seconds, the program will quit.

Having tried this, on my configuration, it doesn't seem to:

/usr/local/astrometry/bin/solve-field --cpulimit 60 -p -D /home/david/solvefield/R/redo/ /home/david/src/R08.fits
/usr/local/astrometry/bin/solve-field --l 60 -p -D /home/david/solvefield/R/redo/ /home/david/src÷R08.fits

...both run past 60s without terminating (in practice, I would increase this limit, but 60s is for testing purposes).

Am I using the syntax incorrectly, or in the wrong order?

Cheers,

David


For the curious, here's my platform:

david@spengler:~$ uname -a
Linux spengler 3.0.0-12-generic #20-Ubuntu SMP Fri Oct 7 14:56:25 UTC 2011 x86_64 x86_64 x86_64 GNU/Linux
david@spengler:~$ cat /etc/lsb-release 
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=11.10
DISTRIB_CODENAME=oneiric
DISTRIB_DESCRIPTION="Ubuntu 11.10"


Dustin Lang

unread,
May 30, 2013, 3:49:25 PM5/30/13
to astro...@googlegroups.com
Hi,

I think cpulimit is only honored correctly if you have the "inparallel" option set in your backend.cfg file.

Without that, if you turn on verbose output, you should see it trying field objects 0-10, 10-20, etc, and I think the limit is per-stage rather than a running total.

Sorry, long-standing bug...

--dstn
Reply all
Reply to author
Forward
0 new messages