[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: Memory consumption in KDE 2.0
From:       Christian Esken <esken () alpha ! tat ! physik ! uni-tuebingen ! de>
Date:       1999-05-03 19:54:24
[Download RAW message or body]

On Mon, 03 Mai 1999 Martin Konold wrote:
>On Mon, 3 May 1999, Uwe Thiem wrote:
>
>> On Mon, 3 May 1999, Martin Konold wrote:
>> 
>> > On Sun, 2 May 1999, Dirk A. Mueller wrote:
>> > 
>> > > Mario Weilguni <mweilguni@sime.com> wrote:
>> > > 
>> > > > As you see, this saver occupies ~6.5 megabytes of data when doing
>> > > > nothing, the saver was activated ~3 times today. I will take a closer 
>> > > > look to the screensavers now.
>> > > 
>> > > Well, why do the screensaves need to hang around in memory if they're
>> > > not just about to show up? Even windows only starts screensaver when
>> > > the timeout has expired and not before. Or is that already done in head
>> > > branch?
>> > 
>> > Well at least with linux the memory occupied by the screensave does not
>> > hurt at all as long as it is not active. The kernel simply throws away the
>> > code pages which are currently not used if the need arises (no swapping or
>> > paging!)
>> 
>> ???
>> 
>> Parse error.
>
>Ok again: The screensaver gets started and then goes to sleep until it
>becomes active. During its long sleeps it stays in memory ONLY if enough

Bla ... I know memory management good enough. Come on Martin, get serious.
Would you want to start every process from inetd.conf just because it might
be needed?

True is, the memory still has to be managed, the process data will
still be there, hogging memory and filling page tables, And the starting
of the screensaver takes time on KDE startup, and so on and so on.

Dirk really has made a really good point here. I would rather have a
screensaver "forker" in memory than the screensaver itself. This needs more
startup time but who cares with a screen saver?




>memory is available. In case the system becomes short on memory the Linux
>kernel will simply throw away those pages which are unaltered and therefor
>can be easily reclaimed from hdd. In case the screensaver is still
>sleeping and memory becomes really short the kernel starts to page dirty
>pages to the /swap partition/file.

Which is rather ugly and could be avoided easily.


>This all means it does not make sense to kill the screensaver and restart
>in when needed.

I see this *very* different. Saving 400K with minimal work is a good thing.
Yes, Martin, I did measurements.

Christian

-- 
Is Unix ready for the desktop? See http://www.kde.org

The                              Christian Esken
|/  Desktop                      KDE Developer
|\  Environment                  esken@kde.org

KDE - The net transparent free Unix Desktop for everyon

[prev in list] [next in list] [prev in thread] [next in thread]