On Jan 13, 3:05 pm, Jason Grout <jason-s...@creativetrax.com> wrote:
> mabshoff wrote:
<SNIP>
> > This is not how this works, i.e. a lot of the memory Sage uses is
> > shared mappings between libraries, so the first notebook process is
> > much, much more expensive than subsequent ones. top and the default
> > interface it uses on Linux is dumb regarding shared mappings, i.e.
> > every notebook process is reported to use every library that is
> > dlopened by Sage.
>
> Okay, I was using top to calculate this. What is a better way to
> calculate the actual memory being used? I'd like to have an accurate
> idea of the memory requirements of a big Sage server.
smap and pmap comes to mind. Some pointers to get started:
*
http://virtualthreads.blogspot.com/2006/02/understanding-memory-usage-on-linux.html
* http://bmaurer.blogspot.com/2006/03/memory-usage-with-smaps.html
The slashdot discussion at
* http://it.slashdot.org/article.pl?sid=06/02/06/120210
also has some lively discussion, but the signal to noise ratio is
pretty bad or normal for slashdot :)
Note that smap has Python binding for Linux, so we might want to
consider using that optionally since not every kernel supports it,
even though on modern system I would guess it will be hard to find one
that doesn't.
In general we are using crappy interfaces to measure memory
consumption from Sage, i.e. get_memory_usage(). We are even parsing
the output from "top" on non-Linux platforms!
> Thanks,
>
> Jason
Cheers,
Michael
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-support@googlegroups.com
To unsubscribe from this group, send email to
sage-support-unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sage-support
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
