Public SourceMage Discussion List

["David C. Haley" <dhaley AT hot.rr.com>]

On Wed, 2004-02-18 at 17:06, Arwed von Merkatz wrote:
> On Wed, Feb 18, 2004 at 03:06:27PM -0600, David C. Haley wrote:
> > On Wed, 2004-02-18 at 13:03, Eric Sandall wrote:
> > > Quoting Wolfgang Scheicher <worf AT sbox.tu-graz.ac.at>:
> > > > As far as i can tell there are less and less problems with 2.6.x 
> > > > kernels.
> > > > I myself switched my work system to 2.6.x when 2.6.2 got released, 
> > > > and with
> > > > my 
> > > > new glibc i burned the bridges behind me.
> > > > 
> > > > So when shall we use 2.6.x as default for the linux spell and the iso?
> > > > 
> > > > We shouldn't do it to early, but propably shouldn't wait too long 
> > > > either.
> > > > 
> > > > I for myself belive that things work well enough to switch test 
> > > > grimoire to
> > > > 2.6.x kernels soon (within the next weeks). And stable somewhen later 
> > > > when 
> > > > it's save. This doesn't mean that users have to upgrade, it's more 
> > > > about what
> > > > 
> > > > shall be default. However, a user should not run into too much 
> > > > troubles when
> > > > 
> > > > using the default :-)
> > > > 
> > > > So - Which kernels are you running, and which problems did you run 
> > > > into 
> > > > because of your kernel?
> > > > 
> > > > Worf
> > > 
> > > All of my machines (except my server) have been using 2.6 for at least 
> > > a few
> > > weeks now (some since 2.6.0-test1 came out), and NPTL since it was 
> > > enabled in
> > > glibc (great work, by the way!). I'd like to see 2.6 as the default 
> > > along with
> > > an ISO (but after 1.0 most likely) as 2.6 fixes quite a few problems 
> > > and is
> > > /really/ fast.
> > 
> > Don't flog me if this is a stupid question. But, since this had started
> > up I've read how people say that it 2.6.x is faster. Is there something
> > out there that can actually measure how fast a kernel functions...or it
> > is just one of those perceptive things?
> 
> It's mostly a matter of perception, as it's not really *faster*, but
> more responsive. If you ever wondered why your windows in X move jumpily
> under a load of 6+, try a 2.6 kernel :)

Never quite had my machine up to a load of six before. I remember /once/
during a programming project I had it up to about 3-3.5 though I can
clearly remember that the memory usage was almost no existant (though
CPU usage was 100%, LOL) and the system went along with no problems at
all. This was back before I was using SMGL though and was still with
2.4.X of some sort. That was honestly teh first time I really thought
Linux was pretty sweet, considering compiling adn running the exact same
program in Windows brought that bloody machine to its knees >:)

Might dig out that code adn recompile now that I have the 2.6.x running
and see if it handles it any differently...

> It definitely seems to be more efficient when your memory usage goes
> through the roof, that could be measurable, though i wouldn't know how
> to do it.

I seem to remember once working on said project above creating a hellish
'little' memory leak. Having almost 2GB or RAM+Swap it ate it up
remarkably quit and though Even teh Linux system to a crawl unlike the
windows counterpart it seemed to work on until I killed it manually.

Not to sure on that...been a while since I've had to use that code for
anything. Though I am interested in seeing how it works now... ^_^

-- 
David C. Haley
dhaley AT hot.rr.com
http://renegadecomputing.net
------------
Source Mage GNU/Linux Developer
General Grimoire Guru
SMGL Security Team Member
------------
PGP Key Fingerprint: 0589 9B39 B259 14D6 82C5 A7E5 AA1B 6596 1F89 0146 
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x1F890146

Attachment: signature.asc
Description: This is a digitally signed message part