Interesting analysis of linux kernel threading by IBM

[Daniel Taylor]

Hate to follow up my own message here, but I realized how it sounds.

What I meant is that if we can optimize for the target case, while noting-
in the source- how to optimize for other cases, this is a good thing.

Hence the comparison with the ide driver, which has (had?) tons of
options for manual optimization in the headers and source.

This way the chosen cases lose nothing, the cases not optimized for lose
little, and the home hacker gains lots.

It is up to others than me to decide what the target case is.

OK, now lets see how crispy I got in the last 6 hours...

Daniel Taylor

On Mon, 24 Jan 2000, Daniel Taylor wrote:

> Just to jump in here:
>   1.5% slowdown on a workstation is insignificant.
>   1.5% speedup on a heavy duty server is quite good.
> 
> If  something like this gets included it would
> be appropriate and desirable to have the
> workstation optimization available from
> the source level (like the IDE driver
> optimizations). This has two positive effects:
> 1. The single-user optimizations are available and
>    can be used by the distributions without custom patches.
> 2. It gives the home "hot-rodder" a hook into learning
>    about the scheduler and an incentive to find it.
> 
> Daniel Taylor
> 
> On Thu, 20 Jan 2000, Giovanni Faglioni wrote:
> 
> > On Thu, 20 Jan 2000, Phillip Ezolt wrote:
> > 
> > > > Hondreds of tasks is just not a typical (perhaps even realistic)
> > > > workload.
> > > 
> > > Yes it is.  
> > 
> > 	Maybe we should let the SysAdmin (user) decide 
> > 	what (s)he's typical workload is. How about a 
> > 	compile-time kernel option ?
> > 
> > 	This will make both the IBM (server benchmarking) 
> > 	folks and the casual home user happy.
> > 
> > 	--Gio'
> > 
> > 
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo en vger.rutgers.edu
> > Please read the FAQ at http://www.tux.org/lkml/
> > 
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo en vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
> 


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/