Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: the strange thing that striked the KC commentator ....about WCCC 2000

Author: Uri Blass

Date: 01:06:57 08/30/00

Go up one level in this thread


On August 30, 2000 at 03:44:07, Tony Werten wrote:

>On August 29, 2000 at 14:44:57, Dann Corbit wrote:
>
>>On August 29, 2000 at 13:13:10, Masciulli Gianluigi wrote:
>>
>>>"...is that during the tournament the programs run on processors of different
>>>speeds. Surely it would be a fairer test of the program if the processor speeds
>>>were equal for all? The programmers seem to think that speed is not all that
>>>crucial (though they typically say it is ‘helpful’), but I suspect it may be a
>>>little more important than they care to admit."
>>>
>>>by J. Levvit at Kasparov chess site.
>>>
>>>I wonder why he can't understand that some programs are build with different
>>>platform in mind and that upgrading from 500 to 750mh cpu is "helpfull" but not
>>>"crucial".
>>
>>Every little bit helps of course.  A faster CPU, faster ram, larger hash, bigger
>>tablebase, better opening book, etc.
>>
>>The fundamental algorithms are most important.  The quality of the opening book
>>is second.  CPU speed and memory are probably next.  Tablebase files are last in
>>importance.
>>
>>Doubling the CPU speed would not be nearly so important as a superior algorithm
>>or bug free opening book.
>
>A very important thing you forgot is the quality of the evaluation. Specially
>passed pawns seem to be important. In the game against SOS, my program
>sacrificed a rook to create 5 (!) passed pawns. One on the 7th, one on the 6th
>and one on the 5th row. 3 out of 5 where on adjacent lines, unfortunately 5 out
>of 5 were useless.
>
>Tony

I think that your program overevaluated the passed pawns.
I believe that reducing the value of the pawns can give your program better
results.

I believe that over evaluating pawns is a common mistake of beginners.
I know that one of the changes in Junior6 relative to Junior5 was reducing the
value of pawns.

Big value of pawns can help your program to score better in test positions
because in these positions there are often sacrifice of a piece for pawns but it
can be also counter productive in games.

Uri



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.