On Sun, 3 Feb 2002, Ed Tomlinson wrote:
> I really think giving the linux schedule more information (not necessarily
> using a shared mm) about which groups of tasks comprise an application would
> help things.
>
> What I coded was an attempt to give the scheduler a way to cope under load.
> If it knows groups of processes belong together then it can control them
> when required. With my current code it place running my freenet node at
> nice +10 still leaves me with a very responsive system.
I don't much like the vm as a way to determine association, but between
process groups and thread groups I would think that if there isn't useful
information perhaps something isn't working as desired.
I firmly believe that nice should be used to get uncommon behaviour, and
that the scheduler should do a decent job with most common loads, which
includes threads.
I hope to try K2 this weekend, I want to get get O1, rmap, and low
latency. If the latest O1 doesn't have child run first I want to add that
as well, barring someone providing a reason not to. I suspect I'll go
head down for some hours doing that, which is why I only evaluate multiple
features every few weeks :-(
-- bill davidsen <davidsen@tmr.com> CTO, TMR Associates, Inc Doing interesting things with little computers since 1979.- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Feb 07 2002 - 21:00:38 EST