From: John Joyce <dangerwillrobinsondanger@...>
Date: 2007-09-25T06:31:28+09:00
Subject: Re: memory use question


On Sep 24, 2007, at 3:57 PM, Roger Pack wrote:

> When I run a program, with these objects, it uses around 15-20MB of  
> RAM,
> thus:
>
> VmPeak:    34308 kB
> VmSize:    31312 kB
> VmLck:         0 kB
> VmHWM:     18808 kB
> VmRSS:     16144 kB
> VmData:    21224 kB
> VmStk:       120 kB
> VmExe:         4 kB
> VmLib:      9320 kB
> VmPTE:        36 kB
> Threads:        2
> ...
> String  8620
> Class   737
> Regexp  410
> Array   327
> Float   312
> Proc    146
> Range   116
> ...
>
> Where Strings are around 2K chars each (I assume 2KB of memory, of
> course that's what I'm trying to figure out).  So I am scratching my
> head wondering if 34MB of RAM is normal for this type of load or not.
> Any thoughts?
> -Roger
> -- 
> Posted via http://www.ruby-forum.com/.
>
Why wouldn't it be?
If you are loading objects into memory and keeping them in use, then  
memory will bloat.
In garbage collected languages, even though memory management is  
technically done for you, the way you write a program still makes it  
bloat or not. But if you're wondering, most new systems sold today  
ship with 1gb of RAM. The OS will generally use all of it that it can  
use and keep some for swapping things in and out in memory.

Garbage Collection is not as sophisticated as the OS's memory  
management for sure!
I'm not claiming to be an elite expert or anything, but the way you  
manage objects even with GC will still determine how big of a  
footprint an app tries to claim.
If you're processing lots of data, you can definitely expect a bigger  
load, but it all depends on how much you try to process  at once and  
what you try to do with it. I see you're using RegExes, and greedy  
RegExes can be big bloaters. But for the same reasons that simply  
loading an entire big text file into objects (one for each line,  
often) can be a big drain too. If you don't know  how big they are,  
or the data source is, you should be limiting it by only reading the  
source of data X number of bytes at a time.

Maybe you already are, but if you're not, you need to manage the flow  
of data in (and out) of the program.

Airports can't process all the passengers at once, so they queue  
them. If planes stop leaving, then people have to stop entering the  
terminals or it will get too crowded to function!