Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Paradise performance

Author: Steven Edwards

Date: 09:40:27 08/22/03

Go up one level in this thread


On August 22, 2003 at 12:26:56, Gian-Carlo Pascutto wrote:

>On August 22, 2003 at 12:19:57, Steven Edwards wrote:
>
>>Paradise was tested on 92 positions picked from the first 100 positions from
>>_Win at Chess_.  (No endgame positions were used.)  It solved all of them,
>>beating out Chess 4.4 (it missed a couple), the monster program of the day way
>>back then.

A correction: the above percentages were only for the first of five groups of
the 92 positions.  For all 92 positions, the performance results were:

Paradise (Wilkins): 97%
Chess 4.4 (Slate and Atkin): 83%
Class A human (Berliner): 83%
Tech (Gillogly): 61%
Caps (Berliner): 59%

Not too shabby program performance for the late 1970s.

>In other words, they cherrypicked the positions it could solve,
>and it solved them (whoa!). You don't mention solution times?

Paradise averaged three minutes thirty-three seconds for each solved position.
This was on a DEC KL-10.  Quite decent for an interpreted language.

>I'm so not impressed.

I've seen the guts of Paradise, and I am impressed.



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.