Author: Bruce Moreland
Date: 00:37:03 02/05/98
Go up one level in this thread
On February 05, 1998 at 02:43:16, JW de Kort wrote: >Hi > >I have written a simple chess program that i want to develope further. >Can anybody recommend me some kind of test that is suited for this task? >Of course i could use some difficult task like the Bratko test, but i do >not think this is very usefull for a weak program. There are a lot of test positions at: ftp://ics.onenet.net/pub/chess/Tests/ A simple tactical test is "wac". This is taken from "Win at Chess", by Fred Reinfeld, published in 1945 (and still in print, as far as I can tell). The suite is easy, once you get your program ironed out you probably get all but 2 or 3 of these right in a minute on fast hardware. Harder ones are "ecm" (Encyclopedia of Chess Middlegames) and "wcsac" (Winning Chess Sacrifices and Combinations). There are some others that purport to be positional tests, as well as some other tactical tests. My advice is to add something that lets you run one of these for however many seconds per position by passing the name of the test to your program via the command line. This way you can make a batch macro that runs several of these while you sleep. It's interesting to run a few of these on each new version you do, so you can chart improvement and catch obvious bugs early. You can use these to try to tune pruning, extensions, move ordering, etc., although you have to take results with a grain of salt. You can also run some of these suites to test out positional terms in your eval function, but in these cases you need to take results with an ocean of salt. bruce
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.