Auto-Adaptive scheduler and semaphore patch ( 2.2.14 ) ...

[David Lang]

On Wed, 26 Jan 2000, Davide Libenzi wrote:

> > 
> > After you can tell where the performance lines cross then you can do a
> > pass 3 that dynamicly switches between the two and see how much that
> > suffers compared to pass 1 (it _will_ suffer becouse you have to run the
> > test to see if you should switch to the new algorithm)`
> 
> It ia done only when scheduler falls into the recharge loop, therefore one
> time on several schedules if the processes are interrupted during their cycle,
> or if they're not interrupted, one time at every 420 ms.
> This is the reason that induce me to thimk that the switch will have a very
> low cost.

i apologize that I haven't gone through the patch but how do you switch
from one to the other? I am assuming that you set a flag that says "use
the old" and at the beginning of the scheduler you test the flag. If this
is the case then that test adds to the overhead (and as tuned as the
scheduler is probably noticably), the other option is to call the
scheduler routine through a pointer, again producing some overhead every
call.

If you hav another method that I have not thought of please mention it.

David Lang


-
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/