Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Does this position blow up your program? 2

Author: Richard Bean

Date: 17:39:42 12/26/02

Go up one level in this thread


>perft 10=950,008,655(1031.73 seconds)

I agree.  As for your program being the fastest to calculate perft, mine
calculated perft 10 from the initial position in a week on a Pentium 4 1.9Ghz -
but first I generated all the unique nodes at a depth of 8, and stored them,
which used up about 37Gb of disk space with my inefficient encoding.  This gave
a speed-up of >85x over not calculating the unique nodes.

With an improvement in my encoding (for example, only storing the difference
between each unique position and the original position) I think perft 11 would
only take a few weeks.  But at some point you have to ask: "What should I do
about 3-time repetition?"  The shortest stalemate problem is more interesting
anyway.

I suppose this "perft calculation" is cheating by some people's definition.



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.