Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Junior 4.6 vs Fritz 5 Match-Game 1

Author: Mark Young

Date: 03:59:21 04/02/98

Go up one level in this thread


On April 02, 1998 at 05:25:48, Harald Faber wrote:

>On April 01, 1998 at 06:17:16, Mark Young wrote:
>
>>>>The programs are running on an autoplayer built into Fritz 5. As far as
>>>>I can tell the programs get 100% of the cpu when its there turn to move.
>>>>This should be as good or better then running the match on two P 200mmx
>>>>computers. So with a 40 move in 3 hour time control this should be a
>>>>fair test. Unless there is some unknown problem with the Fritz 5
>>>>autoplayer that affects the programs in some way.
>>>>
>>>>                                                     Mark Young
>>>
>>>Don't forget the time management in tournament games where permanent
>>>brain together with time management lets programs think between ca.3 and
>>>35mins for some single move!
>>>That is a huge difference I am sure this additional time can't equalize.
>>-----------------------------------
>>
>>You must remember that permanent brain only works if the computer
>>guesses the players next move. I find this only happends about 1 out of
>>every 3 moves.
>
>Huh, then you have seen other games than I did. Rebel, MCP and Genius
>are programs that predict much more moves. Especially MCP profits,
>because when making a move at depth=10 he continues thinking at depth=9
>(of course with the predicted move) where other programs start
>calculating again (also having in mind the opponent's move they await)
>
>>And in computer computer games I have never seen a program do super deep
>>thinks in a game as humans some times do.
>
>Really? Which programs did you use?
>
>> The programs are given a lot
>>of time to do there searches so the best program should still win. If
>>you look at the time stamps on the games that are posted you will see
>>that some of the moves are made after 1000 secs. So if the program need
>>to do a deep think it has the time to do it. Is this match the same as
>>running it on two computers? No but its close.
>
>Hiarcs and MCP are candidates for thinking times of about 20-30mins for
>a move. (in tournament games) Believe me, I saw it with my own eyes,
>they do think for such a long time. Not frequently but it happens more
>often than you think.
---------------------------------------------

Really? Which programs did you use?

I have tested most programs. I just have never seen a program think for
30 min or more on a move at 40 /2hr. That does not mean it can not
happen. I just have not see it.

And a 33% hit ration is about avg. for permenent brain hits. But yes I
have seen it much higher. If the game is mostly tactics you my get over
50% hits.
Some programs seem to be much better at permenent brain hits then
others. I agree Rebel seem to be good at it. Other programs seem to do
less well.



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.