On Wed, 23 Jan 2002, Zdenek Smetana wrote:
> I can't find it.
J5 is the next step towards better interactiveness. Lowered the default
timeslice length from 250 msecs to 150 msecs - long timeslices were
clearly causing problems for certain applications.
there are some changes in my tree that will be -J6, will write a fuller
changelog, there are cleanups from other people included as well.
Ingo
From: Ingo Molnar <[email protected]>
Subject: Re: Missing changelog to Ingo's J5 scheduler?
Date: Wed, 23 Jan 2002 23:28:24 +0100 (CET)
>
> On Wed, 23 Jan 2002, Zdenek Smetana wrote:
>
> > I can't find it.
>
> J5 is the next step towards better interactiveness. Lowered the default
> timeslice length from 250 msecs to 150 msecs - long timeslices were
> clearly causing problems for certain applications.
Yes. -J5 is even better here. With -J4 moving windows arround or doing
other GUI intensive stuff was interactive for a short time (1-2
seconds?) - and then the programm lost all interactivity (with some
unniced gcc in the background ...). With -J5 all applications keep
smoth even with two rebuilds (unniced) of a distribution running!
10:54am up 1:03, 4 users, load average: 4.08, 3.08, 2.39
116 processes: 110 sleeping, 5 running, 1 zombie, 0 stopped
CPU states: 83.6% user, 16.3% system, 0.0% nice, 0.0% idle
Mem: 513912K av, 498908K used, 15004K free, 0K shrd, 144312K buff
Swap: 524624K av, 92K used, 524532K free 256536K cached
> there are some changes in my tree that will be -J6, will write a fuller
> changelog, there are cleanups from other people included as well.
>
> Ingo
k33p h4ck1n6
Ren?
--
Ren? Rebe (Registered Linux user: #248718 <http://counter.li.org>)
eMail: [email protected]
[email protected]
Homepage: http://drocklinux.dyndns.org/rene/
Anyone sending unwanted advertising e-mail to this address will be
charged $25 for network traffic and computing time. By extracting my
address from this message or its header, you agree to these terms.
On Thu, 24 Jan 2002, Rene Rebe wrote:
> Yes. -J5 is even better here. With -J4 moving windows arround or doing
> other GUI intensive stuff was interactive for a short time (1-2
> seconds?) - and then the programm lost all interactivity (with some
> unniced gcc in the background ...). With -J5 all applications keep
> smoth even with two rebuilds (unniced) of a distribution running!
could you also compare -J5 to -J2? [use the 2.4 patch, or vanilla
2.5.3-pre4 which has J2.]
Ingo
From: Ingo Molnar <[email protected]>
Subject: Re: Missing changelog to Ingo's J5 scheduler?
Date: Thu, 24 Jan 2002 12:56:09 +0100 (CET)
>
> On Thu, 24 Jan 2002, Rene Rebe wrote:
>
> > Yes. -J5 is even better here. With -J4 moving windows arround or doing
> > other GUI intensive stuff was interactive for a short time (1-2
> > seconds?) - and then the programm lost all interactivity (with some
> > unniced gcc in the background ...). With -J5 all applications keep
> > smoth even with two rebuilds (unniced) of a distribution running!
>
> could you also compare -J5 to -J2? [use the 2.4 patch, or vanilla
> 2.5.3-pre4 which has J2.]
Ok. After some massive rebooting:
-J2 is worser. starting XFree(+gnome) when three gcc's are running
take long (> half a minute?). With -J5 X start nearly normal (mostly
file access time anyway?) Dragging windows arround is nearly
equal. Although with -J2 i sometimes noticed a really big latency when
starting vim or man ...
Oh. btw. The -J5 was tested with 2.4.18-pre7; the rest was with
vanilla-2.4.17 - I hope this doesn't make a performance difference for
this tests ...
> Ingo
k33p h4ck1n6
Ren?
--
Ren? Rebe (Registered Linux user: #248718 <http://counter.li.org>)
eMail: [email protected]
[email protected]
Homepage: http://drocklinux.dyndns.org/rene/
Anyone sending unwanted advertising e-mail to this address will be
charged $25 for network traffic and computing time. By extracting my
address from this message or its header, you agree to these terms.
On Thu, 24 Jan 2002, Rene Rebe wrote:
> Ok. After some massive rebooting:
>
> -J2 is worser. starting XFree(+gnome) when three gcc's are running
> take long (> half a minute?). With -J5 X start nearly normal (mostly
> file access time anyway?) Dragging windows arround is nearly equal.
> Although with -J2 i sometimes noticed a really big latency when
> starting vim or man ...
thanks, so -J5 is an improvement on all fronts - good.
> Oh. btw. The -J5 was tested with 2.4.18-pre7; the rest was with
> vanilla-2.4.17 - I hope this doesn't make a performance difference for
> this tests ...
i dont think there is a difference, as long as you have enough RAM and
dont swap usually.
Ingo
In mlist.linux.kernel, you wrote:
>
> J5 is the next step towards better interactiveness. Lowered the default
> timeslice length from 250 msecs to 150 msecs - long timeslices were
> clearly causing problems for certain applications.
>
> there are some changes in my tree that will be -J6, will write a fuller
> changelog, there are cleanups from other people included as well.
I'm running now J5 on 2.4.17 vanilla, UP.
Everything runs smooth. No problems whatsoever.