Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Tao at ICC

Author: Bas Hamstra

Date: 09:26:00 01/18/02

Go up one level in this thread


On January 18, 2002 at 02:47:27, Tony Werten wrote:

>On January 17, 2002 at 11:37:49, Bas Hamstra wrote:
>
>>On January 17, 2002 at 11:19:40, Gian-Carlo Pascutto wrote:
>>
>>>On January 17, 2002 at 11:18:05, Bas Hamstra wrote:
>>>
>>>>Played Leila/Crafty/Yaces/Tigers/Fritz/Hossa 15/0 and 30/0 at ICC:
>>>>
>>>>statistics for TaoYin(C)         On for:    0     Idle:    0
>>>>
>>>>          rating [need] win  loss  draw total   best
>>>>Standard    2534         12     5     3    20   2534 (17-Jan-2002)
>>>>
>>>> 1: Chess program Tao 5.00
>>>> 2: Athlon 1.4
>>>> 3: Bas Hamstra (bas.hamstra@wxs.nl)
>>>>
>>>>At these rapid tc's it does ok so far. Hardware was mostly comparable (only
>>>>Hossa on a 800 Mhz machine). At first sight it seems to do not bad at all
>>>>against Crafties and Yaces on equal hardware. It might very well be totally
>>>>different at longer time controls, no idea.
>>>
>>>Are you doing checks in the quiescent search again? :)
>>>
>>>--
>>>GCP
>>
>>Well, yes :-) Since I got used to that years ago, I cannot understand how the
>>others do so well without it. For me the whole "mini-qsearch with futility
>>pruning" concept is one big myth. At some points I actually agree with Vincent.
>
>:)
>
>Have a question that appears to me every time the question "Do you do checks" >raised. Does this mean you force the oppo to search all moves when he's in
>check or do you search checkgiving moves as well in quies ?

You can do only "evasions" or you can do checkgiving moves, in which case you
must also do evasions. I do checkgiving moves. The worst IMO is doing neither
together with FP pruning in the qsearch. It misses so much that I think the
depth you gain by this is mostly "optical". Pruning SEE<0 is excellent, but FP
sucks.

Compare FP with lazy eval handling things *accurately*. When a FP cut is
"right", lazy eval would have cut it also, and the only thing you gain is a
recursive call to quiesce() and a lazy eval. That won't buy you a lot of depth.

Best regards!
Bas.



















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.