Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Endgame Test Suite - needs correction

Author: John Hatcher

Date: 21:15:52 08/04/01

Go up one level in this thread


On August 04, 2001 at 20:17:10, Peter McKenzie wrote:

>Is anyone aware of a good endgame test suite?
>
>I am after a test suite that has the following properties:
>
>- a decent number of positions, I'd say at least 100 positions.
>
>- a range of different endgame types: pawn endings, minor piece endings, rook
>endings, queen endings, multi-piece endings.
>
>- positions that cover a range of difficulties, a few easy ones and quite a few
>hard ones.  A reasonable program should be able to solve some at say
>10sec/position, but it should not be too easy.  If programs could solve say 95%
>of positions at 1min/position, I'd call it too easy.
>
>- the test suite itself should be thoroughly debugged!
>
>
>If such a suite does not exist, perhaps it would be a good project to create
>one?  A good start might be trawling back thru. CCC for test positions.  Is
>there much interest in this idea?
>
>Peter

A quick check of the test suite solutions using 5 piece tablebases shows that
several of the positions have either:

1.  a solution which is not the best/quickest
2.  other solutions which are exactly equivalent

A incomplete list of the problem positions include #18, #58, #63, #70, #81, #98,
and #99.  I'm dubious of a few others - especially #29 - but don't have time to
thoroughly test it.  Maybe someone else has the time to debug this suite.

JOHN




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.