Author: Christophe Theron
Date: 03:12:26 08/18/98
Go up one level in this thread
On August 12, 1998 at 21:51:31, Hans Christian Lykke wrote:
>On August 12, 1998 at 09:31:53, Christophe Theron wrote:
>
>>On August 12, 1998 at 08:02:07, Hans Christian Lykke wrote:
>>
>>>Hej!
>>>
>>>I have started playing the Nunn-test on my P200MMX 64MB, between the
>>>engines Hiarsc 6 and Fritz 5.0. (both get 23MB hash)
>>>
>>>I will play it 3 times with the following levels.
>>>
>>> 67 seconds per move.
>>>200 seconds per move.
>>>600 seconds per move.
>>>
>>>How will this end up?
>>>
>>>What does all the "wise" guys on CCC think?
>>>
>>>Let´s have some predictions.
>>>
>>>
>>>Venlig hilsen
>>>
>>>Hans Chr. Lykke
>>
>>I think the x seconds per move setting is a mistake. If you sum up the
>>computation times once the match is finished you could well realize that one
>>program has taken much more time that the other one.
>
>My experience playing with ex. 1 hour with engine/engine is, that in too many
>games one of the programs will loose on time!
>I know that loosing on time is a part of chess, but I don´t think that is so
>good in this test.
>
>Here are the total times of some of the games played until know
>
>1:20:51 1:20:31
>1:07:34 1:07:37
>0:45:29 0:45:53
>7:30:21 7:42:24
>1:19:47 1:10:46 0-1
>5:26:01 3:49:19 0-1
>
>So I think you are not right that one program has taken much more time that
>the other one.
>
>In the two last games on the list there is a greater difference, but the program
>using most time, lost!
>
>
>Venlig hilsen
>
>Hans Chr. Lykke
Good to see that you check this kind of things.
Why don't you use "x moves in y minutes"? It would avoid the sudden death
problem...
I don't like your x sec per move because of the experience I have with Chess
Tiger. I did not work very long on the x sec per move level, because it is not
used in real tournaments. I took a long time working on the x moves in y minutes
and on the sudden death time controls, because it is used in tournaments, and it
can make a real difference if it is well programmed.
If you use Chess Tiger in the future, I would strongly advice NOT to use the x
sec per move level, because the program is much stronger when it is allowed to
manage its own time along the entire game.
The time allocation algorithm is part of the strengths of the modern programs,
and you just disable it with your x sec per move.
Christophe
>
>
>
>>
>>You should use standard time controls (x minutes for y moves) or sudden death
>>time controls (x minutes for the game) so that the time problem keeps under
>>control. For the levels you have given, you could for example take:
>>
>>1) 1 hour for the game
>>2) 3 hours 30 minutes for the game
>>3) 10 hours for the game
>>
>>
>> Christophe
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.