Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A question for winboard experts

Author: Uri Blass

Date: 04:17:27 07/05/03

Go up one level in this thread


On July 05, 2003 at 07:05:17, Sune Fischer wrote:

>On July 05, 2003 at 06:52:29, Günther Simon wrote:
>
>>On July 05, 2003 at 06:25:20, Frederic Louguet wrote:
>>
>>>>Do you have the feature done=1 in your program?
>>>>Take a look at the link below, where someone asked
>>>>a similar question in the Winboard forum:
>>>>http://f11.parsimony.net/forum16635/messages/50687.htm
>>>
>>>With "done=1", the program starts faster the first time, yes. However, Winboard
>>>itself takes 10 seconds between each game in a n-games match, and this is really
>>>annoying :)
>>
>>Well this gives a delay of around 17' after 100 games ;-)
>
>If these 100 games are at 5 sec per game this delay doubles the tournament time.
>
>>Have you ever tried Joris' WBTM (WinboardTourneyManager)?
>>I use this almost exclusively, because I can run round robin
>>and swiss tourneys too, moreover I can save the debug files
>>one after one not in a single file only.
>>I am talking about WBTM because there are some options
>>in its config like:
>>
>>#    Runner interval in milliseconds. Runner will see every N milliseconds if
>>game is finished or another is ready to start.
>>     RunnerInterval         30000
>>
>>#    ScheduleInterval in milliseconds. Schedule will see every N milliseconds if
>>runner state should be switched.
>>     ScheduleInterval       30000
>>
>>BTW as you might have thought, it creates batch files which start the
>>created games, may be if you set those intervals to a very low number
>>you get what you want, as WB is always new started and you say your
>>program starts fast in the first game :)
>
>If winboard is started over it's no good.
>Restarting the program takes even longer, you have to free and reinitialize the
>hash (~ 1 sec for any decent size hash), and you don't get the level of testing
>you want because you don't catch the bugs there might be in not restating the
>program.

I agree with the second part that you cannot catch bugs that happen when you do
not restart the program.
I do not agree that restarting the program takes even longer because of hash.
programs do not need big hash in this kind of test and 1 Mbyte is clearly
enough.

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.