Re: where does the memory go?

[Matt Nelson <mnelson dynatec com>]

On Tue, 9 Mar 1999, Daniel J. Frasnelli wrote:
> And let's compare this to my system with 128M ram:
> [dfrasnel@relativity /tmp]$ free
>              total       used       free     shared    buffers     cached
> Mem:        127312     115416      11896      73936      13832      61216
> -/+ buffers/cache:      40368      86944
> Swap:       131064        632     130432
> 
> If you look at both of these 'free' outputs, you will notice that a good 
> amount of memory is tied up in both cached and shared memory allocation.
> My system shows approx. 11M of physical memory free.  However, if I were to 
> launch Netscape 4.51 (memory footprint of what, 30M?) I could do so
> without dipping into swap space.  

what you say is true, but i was more concerned with the -/+ buffers/cache
line of the free(1) output.  that line, if i understand it correctly,
indicates what your are using and have available if you dont count buffers
and cache.  in your example, it shows that you have about 40M used and
about 87M free.  *those* are the numbers that i'm wondering about.  on my
machine, the -/+ buffers/cache used climbs to ~110M during the activity of
the day, then drops down to ~25M during the night.

when i compare the 110M to the output of "ps axm" during a quiet time in
the day (i.e., no user jobs running), i see that the the total memory used
doesn't seem to add up to the 110M that free(1) indicates.  and yes, X has
the biggest memory footprint on an idle machine...

at first i thought there was a memory leak somewhere, but when it went
back down to something reasonable over night, i figure that i must not be
understanding something.

any other ideas? 

-matt

-----------------------------------------------------------------------------
Matthew Nelson
Dynamics Technology, Inc. 
21311 Hawthorne Blvd., Suite 300, Torrance, CA 90503-5610
Voice: (310) 543-5433   FAX: (310) 543-2117   Email: mnelson@dynatec.com