Re: 2.4.22aa1

[Andrea Arcangeli]

On Tue, Sep 02, 2003 at 08:38:39PM +0200, Marc-Christian Petersen wrote:
> On Tuesday 02 September 2003 04:02, Andrea Arcangeli wrote:
> 
> Hi Andrea,
> 
> > Only in 2.4.22aa1: 20_sched-o1-fixes-10
> > Only in 2.4.22pre7aa1: 20_sched-o1-fixes-9
> > 	Changed the CHILD_PENALTY logic to be centered around
> > 	50%. From Kurt Garloff.
> 
> the changes 's/CHILD_PENALTY/CHILD_INHERITANCE' and "s/PARENT_PENALTIY//' are 
> really awfull for desktops. If I change child_inheritance from 60 to 95 and 
> reintroduce the logic with parent_penaltiy, it's alot smooter under load.
> 
> I think these logics should be #ifdef'ed with CONFIG_DESKTOP, no?

s/PARENT_PENALTIY// is a noop. The CHILD_INHERITANCE moves it from 50%
to 60% and center it in the middle. If something it will increase the
sleep_avg of the child that will lead to making the childs more
responsive. It sounds quite strange that backing out this, makes the
system more responsive. btw, Kurt developed this stuff with the object
of making it more responsive, not less. What could go wrong is that the
child will get more responsiveness than the parent and maybe for your
load the interactive app is the parent and not the child, dunno.

Andrea

/*
 * If you refuse to depend on closed software for a critical
 * part of your business, these links may be useful:
 *
 * rsync.kernel.org::pub/scm/linux/kernel/bkcvs/linux-2.5/
 * rsync.kernel.org::pub/scm/linux/kernel/bkcvs/linux-2.4/
 * http://www.cobite.com/cvsps/
 *
 * svn://svn.kernel.org/linux-2.6/trunk
 * svn://svn.kernel.org/linux-2.4/trunk
 */
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/
On Tue, Sep 02, 2003 at 08:38:39PM +0200, Marc-Christian Petersen wrote:
> On Tuesday 02 September 2003 04:02, Andrea Arcangeli wrote:
> 
> Hi Andrea,
> 
> > Only in 2.4.22aa1: 20_sched-o1-fixes-10
> > Only in 2.4.22pre7aa1: 20_sched-o1-fixes-9
> > 	Changed the CHILD_PENALTY logic to be centered around
> > 	50%. From Kurt Garloff.
> 
> the changes 's/CHILD_PENALTY/CHILD_INHERITANCE' and "s/PARENT_PENALTIY//' are 
> really awfull for desktops. If I change child_inheritance from 60 to 95 and 
> reintroduce the logic with parent_penaltiy, it's alot smooter under load.
> 
> I think these logics should be #ifdef'ed with CONFIG_DESKTOP, no?

s/PARENT_PENALTIY// is a noop. The CHILD_INHERITANCE moves it from 50%
to 60% and center it in the middle. If something it will increase the
sleep_avg of the child that will lead to making the childs more
responsive. It sounds quite strange that backing out this, makes the
system more responsive. btw, Kurt developed this stuff with the object
of making it more responsive, not less. What could go wrong is that the
child will get more responsiveness than the parent and maybe for your
load the interactive app is the parent and not the child, dunno.

Andrea

/*
 * If you refuse to depend on closed software for a critical
 * part of your business, these links may be useful:
 *
 * rsync.kernel.org::pub/scm/linux/kernel/bkcvs/linux-2.5/
 * rsync.kernel.org::pub/scm/linux/kernel/bkcvs/linux-2.4/
 * http://www.cobite.com/cvsps/
 *
 * svn://svn.kernel.org/linux-2.6/trunk
 * svn://svn.kernel.org/linux-2.4/trunk
 */
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/