Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is crafty the fastest program to calculate perft with no hash tables?

Author: Heiner Marxen

Date: 16:06:53 06/20/01

Go up one level in this thread


On June 20, 2001 at 18:41:59, Peter McKenzie wrote:

>On June 20, 2001 at 16:06:34, Dann Corbit wrote:
>
>>On June 20, 2001 at 16:00:53, Peter McKenzie wrote:
>>[snip]
>>>Oh, btw, LambChop can count the 6ply tree from the root in 20sec on PIII
>>>1000mhz.  It is faster than crafty because it counts moves at depth 5 not
>>>positions at depth 6 and hence effectively it has to search one ply less in the
>>>tree.  It can do this because I have implemented a fully legal static move
>>>generator that does not need makemove to exclude illegal moves.
>>
>>!!!
>>How did you do that?
>>Smoke and mirrors?  Lookup tables?  Plugged into a cray?
>>!!!

No smoke & no mirrors, Dann!


>Its not so hard if you compute attack information already.  Plus I keep track of
>pins too which helps.
>Its not necessarily the best way, but it does have certain advantages and its
>fun which is reason enough :-)
>
>Peter

Yes, for Chest it is quite similar, and the timing is also similar:
  56.440 sec for perft 6 on K6-3/400.
No magic involved, just efficient generation of fully legal moves.

Heiner



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.