Re: new kernel 2.4.20-13.9 performance improvements?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, 2003-05-19 at 05:03, Klaasjan Brand wrote:
> 
> I tried to 2.4.20-13.9 kernel after hearing about performance
> improvements for interactive use. On both systems I use regularly I

Ok, i'm running the new kernel update on 8.0, not on 9, so if that's not
relevant, please dismiss this message.

Anyway, this desktop is actually a server hardware - dual PIII/800,
768MB RAM, 4 x SCSI drives (9 and 18GB), crappy graphics, PS/2 mouse and
keyb.

The new kernel is obviously faster for disk read operations. The Gnome
menu icons simply pop up immediately; before, there was a visible delay
in loading them. The desktop comes up faster when i log in to the
graphical interface.

OTOH, Galeon crashed yesterday. I didn't see that happening at all in
the last year or so, not when running it under Gnome anyway.

And... i'm running a bunch of MRTG stuff from cron every 5 minutes. With
the previous kernel (i think it was the previous 8.0 kernel update) and
with all the older 8.0 kernels, firing up the whole bunch of MRTGs was
no issue, which i considered to be normal (hey, it's a SMP system :-D).
The MRTGs stole the whole CPU cycles available on both CPUs, but that
was ok, i didn't see any negative effect on the desktop.
With the latest kernel update though, when the MRTG gang goes out
trigger-happy (which happens every 5 minutes and has a duration of 10
seconds or so), the mouse stutters, the desktop apps are sluggish.
I don't see any interactivity improvements here, it's actually quite the
opposite.

To me, it seems like this kernel has been heavily optimized for servers.
That's what Red Hat seems to be usually doing, or so i think, but this
time it looks like they pushed that a lot further.
However, at least in my case, the desktop stuff took a hit.

I'm still reluctant to upgrade my RH 9 system (my other desktop) to the
latest kernel, due to lots of custom kernel modules i'm using there
(ALSA, nForce, nVidia, LIRC, etc). If/when i do that, i'll repeat the
same tests and see how that goes. Maybe version 9 is somewhat different.

> experienced some "stutter" when loading big applications (evolution,
> mozilla) which look to me like they're caused by heavy disk activity.

Like i said, in my case the stuttering is caused by a bunch of Perl
scripts going off. No big disk activity, just some 10...20 processes
eating up all CPU cycles.
That didn't happen with previous kernels.
OTOH, due to the fact that i have 4 SCSI disks on my RH 8.0 system, and
the filesystem is cleverly spread out across them (major I/O hogs are
each on it's own spindle), i only rarely have issues with disk I/O on
that system, no matter what.

At least in the case of my RH 8.0 system, the problem with the latest
kernel update seems to be the CPU usage, not disk I/O.

I'm afraid to think what's going to happen with my RH 9 system if i
upgrade it to the latest kernel and then start using the CPU hogs like i
usually do (heavy multimedia stuff). With the previous kernel, that
system works exceptionally well (as a desktop) under high CPU usage.

> On my work system, a P4/i850 system  the new kernel seems to perform a
> lot better. Random "pauses" don't occur anymore.

Yeah, the disk stuff seems smoother.

-- 
Florin Andrei

"Good people do not need laws to tell them to act responsibly,
while bad people will find a way around the laws." - Plato




[Index of Archives]     [Fedora Users]     [Centos Users]     [Kernel Development]     [Red Hat Install]     [Red Hat Watch]     [Red Hat Development]     [Red Hat Phoebe Beta]     [Yosemite Forum]     [Fedora Discussion]     [Gimp]     [Stuff]     [Yosemite News]

  Powered by Linux