Author: Matthew Hull
Date: 22:16:44 02/16/06
Go up one level in this thread
On February 17, 2006 at 00:54:17, Tom Dyer wrote:
>On February 16, 2006 at 23:17:30, Robert Hyatt wrote:
>
>>On February 16, 2006 at 22:09:11, Tom Dyer wrote:
>>
>>>Hi -
>>>
>>>I am running an experiment with Crafty 20.1 (Bryan Hoffman W32 compile) of
>>>Crafty with Vs Crafty without the complete 3-4-5 egtbs. Running 1000 games per
>>>match. In my both my calibration match (Crafty egtbs enabled vs Crafty egtbs
>>>enabled) and my test match (Crafty EGTB vs Crafty No EGTB) everything seemed
>>>fine until game 298 and 299. Crafty stopped logging the game.### and log.###,
>>>and keeps dumping the contents of the latest game into 299.*. The
>>>
>>>Here is what my folders look like:
>>>
>>>...
>>>log.297 46KB 297 File 2/14/2006 11:13am
>>>log.298 60KB 298 File
>>
>>This is normal. I don't let the log file go past 299. You can look for this in
>>init.c and change it. I just wanted to avoid leaving thousands of the files
>>around when I forgot to clean them out. :)
>
>Thanks! One other question; in my calibration match (Crafty w/3-4-5 tb Vs
>Crafty w/3-4-5), everything seemed to be going along fine (unique games
>played)until games 477-527 - it kept repeating the same game. There were a few
>other common games I also saw repeated throughout the match. Any idea what
>could have caused this behavior? Here is one of the games that kept repeating:
>
>[Event "Computer chess game"]
>[Site "TOMSOFFICE"]
>[Date "2006.02.11"]
>[Round "485"]
>[White "Crafty-20.1 BH"]
>[Black "Crfty2-20.1 BH"]
>[Result "1-0"]
>[TimeControl "60+3"]
>
>1. d4 d5 2. c4 dxc4 3. e3 Nf6 4. Bxc4 e6 5. Nf3 Bb4+ 6. Bd2 Be7 7. O-O O-O
>8. Nc3 Nc6 9. Rc1 Rb8 10. Qe2 Ra8 11. Bb5 Bd7 12. e4 Rc8 13. Bc4 Nb4 14. a3
>Nc6 15. e5 Ng4 16. Qe4 h5 17. h3 f5 18. Qe2 Nh6 19. Bxh6 gxh6 20. d5 Na5
>21. dxe6 Bc6 22. Rfd1 Qe8 23. Bd5 a6 24. b4 Bxd5 25. Nxd5 Nc6 26. Qe3 f4
>27. Qe4 Bd8 28. Nxf4 Ne7 29. Nh4 c6 30. Nhg6 Nxg6 31. Nxg6 Qxe6 32. Nxf8
>Kxf8 33. Rd6 Qf7 34. Rxh6 Kg8 35. Rg6+ Kf8 36. e6 Qh7 37. Rf6+
>{Black resigns} 1-0
>
>Here is crafty.rc file: I gave 1 GB RAM and all programs (except Crafty are
>de-activated on the machine while the experiment is running)
>
>hash=192M
>hashp=48M
>cache=32M
>xboard
>computer
>log on
>ponder on
>drawscore=0
>learn 7
>book on
>book random 1
>book width 5
>resign 0
>level 0 1 3
>tbpath=c:\Progra~1\Crafty\tb
>egtb
>
>2nd instance of Crafty is called crfty2 and the only difference in the rc is the
>path to the tb files.
From my experience, this can happen when the book is not wide and book learning
enabled. The program can get into a rut where it prunes unsuccessful lines, but
starts running out of alternatives lines to play. If you are going to play 1000
games with book learning, start with a wide book. If you have a super-narrow
tournament book, it can get you into a repitition rut really fast after about 30
games or even less. But even with a wide book, you may end up with duplicate
games after 1000. Once your experiment is complete, you'll want to toss dupes
from the sample.
Matt
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.