Re: Overscheduling DOES happen with high web server load.

Andrea Arcangeli (andrea@e-mind.com)
Fri, 7 May 1999 00:06:41 +0200 (CEST)


On Thu, 6 May 1999, Phillip Ezolt wrote:

>On Thu, 6 May 1999, Andrea Arcangeli wrote:
>
>> On Wed, 5 May 1999, Phillip Ezolt wrote:
>>
>> Alt. Alpha uses HZ at 1024 so you get a scheduling rate by default 10
>> times higher than in all other archs.
>
>This is interesting. Is there any reason why alpha is at 1024?

I only used Alpha one time in my life for some hour so I don't know ;). I
guess it's because Alpha is (was) faster.

>I changed your patch slightly, and ran the test again.

Really it seems to me that you have not changed it but you have simply
applyed it by hand ;).

>In any event, there is an HUGE amount of overscheduling going on.

I was worried about that infact (I also tried to fix that in the 2.1.x
cycle but somebody complained when I replaced the `if (->counter + 3 <
xxx->count) with `if (goodness(xxx) < goodness(xx))' telling me that the
current reschedule_idle() was just smart enough...).

I'll provide you a patch shortly to try out.

> procs memory swap io system cpu
> r b w swpd free buff cache si so bi bo in cs us sy id
>
> 0 0 0 0 226872 1544 9816 0 0 0 0 1099 39056 2 2 96
^^^^^
amazing

>Is there any penalty for doing this?

Yes: you are wasting time 39056 times/sec in the scheduler.

Andrea Arcangeli

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