Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty log files

Author: James T. Walker

Date: 17:59:56 03/05/99

Go up one level in this thread


On March 05, 1999 at 19:01:49, José de Jesús García Ruvalcaba wrote:

>On March 05, 1999 at 18:56:05, James T. Walker wrote:
>
>>On March 05, 1999 at 15:31:02, Robert Hyatt wrote:
>>
>>>On March 04, 1999 at 22:58:30, James T. Walker wrote:
>>>
>>>>On March 04, 1999 at 21:15:35, Jeremiah Penery wrote:
>>>>
>>>>>On March 04, 1999 at 18:00:12, James T. Walker wrote:
>>>>>
>>>>>>Hello,
>>>>>>Is it possible to determine if Crafty exceeded the time limit by looking at the
>>>>>>log files??
>>>>>>Jim Walker
>>>>>
>>>>>If Crafty is losing on time there must be something wrong with the
>>>>>interface/adapter it's using.  For example, Crafty once played 257 moves in a 1
>>>>>minute game before _I_ finally ran out of time (I was just shuffling pieces back
>>>>>and forth :).  There are certain settings that could cause it to lose on time,
>>>>>but I can't imagine it doing so except in very fast games.
>>>>>
>>>>>Jeremiah
>>>>
>>>>Hello Jeremiah,
>>>>Well I'm certain it is losing on time.  I'm using version wcrafty16.5.  I've
>>>>noticed it stops printing the board after each move around move 68-70.  Later it
>>>>stops printing the PV's and just prints the move.  I don't know is this is done
>>>>to conserve time or what.  It also starts giving iterative deepening errors in
>>>>the log near the end also.  I suspect all of this is a bug.  I'm going back to
>>>>version 16.2 and see if it does the same thing.
>>>>Jim Walker
>>>
>>>
>>>The program does 'throttle' log output when low on time, for efficiency
>>>reasons (particularly 1 0 games on ICC).  Ie it doesn't display the board,
>>>nor log any analysis once the target time is under a second per move...  But
>>>it can play 100 move games in a single second, so it can handle that with no
>>>problem...
>>
>>Yes,  this all makes sense to me now.  I wrote this before I had your info
>>regarding crafty and time control.  But I did go back to Crafty 16.2 and found
>>that it misplayed a endgame using <EGTB> so I just scrapped all the 16.2 games
>>and also the 16.4 games and I'm starting over with my test.  I've never seen
>>such strange results as I'm getting now.  Sometimes when I start a series of
>>games Crafty seems to give Junior and Fritz fits.  Then the next day I start
>>anoter series and Crafty gets blown away.  It's like it has two different
>>personalities.  All I can do is combine all the results and see what happens.
>>Jim Walker
>
>If you are using compressed tablebases (*.emd files) then you need crafty 16.5,
>as previous versions only understand uncompressed ones.


No, I'm not using the compressed TB's.  I'm considering going to them now but
hate to download all the TB's again.  I can't get the 5 man TB's because AOL
keeps cutting me off.
Jim Walker



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.