Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: perft records

Author: Reinhard Scharnagl

Date: 11:29:43 09/06/04

Go up one level in this thread


On September 06, 2004 at 14:18:47, Pallav Nawani wrote:

>On September 06, 2004 at 09:06:06, Reinhard Scharnagl wrote:
>
>>On September 06, 2004 at 07:43:51, Peter Fendrich wrote:
>>
>>>What free programs have the fastest perft and what are the figures?
>>>Please, If you give figures also add processor, compiler and environment!
>>>
>>>I want to compare with a new concept that isn't coded yet...
>>>/Peter
>>
>>Hi Peter,
>>
>>see some of my (Smirf) results (still improved after measuring) at:
>>
>>a) without TT: [http://www.chessbox.de/Down/CRC_Test_03.txt]
>>b) with    TT: [http://www.chessbox.de/Down/CRC_Test_04.txt]
>>
>>Regards, Reinhard.


>Hi,

Hello,

>Two questions:
>1. How do you use TT in perft?

I store the countings for an individual position and its level into the TT.
Because of the problem to store different values for a unic situation you
have to make their keys different for each level.

>2. What's the idea behind using TT in perft? I thought perft is just used to
>check the accuracy of move make/unmake & speed of raw move make/unmake.

With normal perft I want to verify the ability of my data structures to build
up a performant chess engine and to generate correct moves. (For that goal I
have provided a differrentiated statistic of move types, which gave me the
chance to detect errors very precisely. In a pseudolegal move generator the
chances to have bugs in the generator are very low compared to a fully informed
legal move generator like in Smirf.

With a perft using TTs I try to verify the efficiency of caching and the answer
of the question whether I would generate hash keys correctly. Using TT perft I
found indeed some bugs and could correct the code. It would be good to know,
how the performance of perft could be improved by other engines using TT, for
to investigate, whether the usage quality of the TT is average, better ore
worse.

>Regds,
>Pallav

Regards, Reinhard.



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.