Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Interesting statement from Ossi Weiner about Nunn test

Author: Serge Desmarais

Date: 15:05:41 09/18/98

Go up one level in this thread


On September 18, 1998 at 14:01:43, Bruce Moreland wrote:

>
>On September 17, 1998 at 21:32:58, Serge Desmarais wrote:
>
>>   You seem to give too much importance to the engine difference... I think it
>>was said that Fritz on a K6 200MMX (as opposed to a P II 200MMX) was about
>>25-30% slower MAXIMUM. But to reach a depth of ONE ply deeper in the same time,
>>one computer has to be between 500 to 600% faster!
>
>This is an old argument that I do not agree with.  The false premise is that in
>order to get stronger you have to search another ply.
>
>Imagine it this way instead.  Rather than saying, 25% is useless because it
>doesn't give you an extra ply, consider thatn with a 25% increase in nodes per
>second, a shot that takes you a minute to find before will now only take 80% of
>this time to find, which is 48 seconds.
>
>I would gladly suffer a lot in order to find a 60-second shot in 48 seconds.
>
>I think that what I have said is enough, but I think that I can continue and
>find another flaw.  If it is fair to give one side a 25% disadvantage, it should
>be fair to give a 25% advantage instead.  Since either of these is fair, perhaps
>you could randomly pick between these alternatives with no concern.
>
>Choice one is to let you go 80K nps while your opponent goes 100K nps, which
>means you are 25% slower.  Choice two is to go 125K nps while your opponent goes
>100K nps, so you are 25% faster.  If you think for me to go 25% slower is fair,
>then you should have no problem with my going 25% faster.  But the difference
>from my point of view is 56%.  Whatever Newborn says, I will win more games if I
>go 56% faster than I otherwise would.
>
>Both of the extremes cannot be "fair" at the same time, given your lower bound
>of 25%.  You can't just look at the deviation from equal hardware, you have to
>consider what would happen if the hardware were swapped, in order to really see
>it from your opponent's point of view.
>
>bruce


   Of course, the faster you can go, the better! But I said the above because
Youri said that a won match could become a lost match, by decreasing the speed
of a program by 25%. If a program won a match by 21-19, even with no decreasing
in spead, it could lose the next one. Just that I don't think that if the same
program won 30-10, it could lose only because a 25% decrease in speed?


Serge Desmarais



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.