Author: William Kerr
Date: 07:59:43 07/31/99
Ive written a chess program which I started in 1993 and modified and improved it over the years until it plays at the Neanderthal level. It is a depth first, alpha-beta, and killer-heuristic program with positional knowledge taken from an article authored by the Chess 0.4 (remember Chess 4.5) team which appeared in Byte Magazine several years ago. My program is called WKchess and has beaten a commercial chess program (I beleive the program was called "Grandmaster chess". It played GNU chess once and GNU chess crushed my program in in the end game. WKchess is written in C++ (actually C code). From memory I used the move generation algorithms from Dan and Kathe Sprakelin (sp). However the program generates well over 550,000 nodes/second. If there is any thing good about my chess program is thats its exceeding well commented (so I can remember how it works) and easy to understand since I used no tricky or confusing 'C' programming tricks. Enough of plugging my chess program, what I noticed is that Borland's C++ 4.52 runs 32 bit integer code (no floating point) about twice as fast as the learning edition of MVC C++ Ver 6.0 as supplied in the Learn C++ Programming from Sams books. However, the floating point speed of the two compilers is exactly the same. I would assume the professional version of VC Ver 6.0 with optimizations turned on is as fast as Borlands C++ compiler. Beware the execution speed of various C++ compilers can vary by quite a bit. Bill
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.