Author: Peter McKenzie
Date: 18:10:47 05/29/01
Go up one level in this thread
On May 29, 2001 at 13:40:22, Frank Phillips wrote: <snip> > >TSCP certainly helped enormously. A lot of stuff is in one place, visible and >its connection with other stuff apparent, and as you say there are also ready >made answers to several questions including how to record castling status and so >forth, which introduced me to new areas and the nitty gritty of implementation. > >By this time I got the courage to try chess, I had tried other more basic things >(did not seem basic then of course, but now look primitive even to me!), knight >tour, connect4, simple othello, simple draughts etc I still have some of the >code labelled for example, connect4 another recursion project :-) So I did >not start with tscp and hack it around, I actually started with a blank sheet >(.c file). The first attempt was pretty similar, although it did not stop at 4 >ply. My model was Marsland's recipe, with tscp demonstrating how (mostly the >non search stuff I guess) might be implemented in practice. Yes saved me at >least a year and probably from giving up. GNU chess code was way over my head. >You probably need to have written a program before you can understand it, >because to analyse it you need the basic structural elements in your head? >Crafty....well I just read the comments, which are an insight in themselves. >The code tended to be a deterrant. > >Now my program is basically bitboards with precomputed offset arrays for >sliders. I flirted with 0x88 for a short time before deciding on bitboards, >mainly because I believed they would make evaluation easier. I would say this is >my second, but in reality _my_ first program, if you see what I mean. > >I keep meaning to start again with a clean sheet and perhaps 0x88 to try to >discover the advantages this method has. Bruce and Christophe seem to like it. >How many times have you started with a completely sheet (.c file)? I've started just once, although at one point I did write an 0x88 version of my program but threw it away for some reason. Oh, I did try to write a chess program in Apple II Basic when I was about 14 but failed miserably :-) I wrote my thing from scratch in about 1992 or 93 and had it working before I took a look at the GNU Chess source which I didn't understand and so I quickly gave up on it. I do remember being confused about how to calculate a hash key, so I downloaded the Arasan code and that provided me with the answer (thanks Jon :-). I've never studied the crafty code, but I have had alot of help from Bob over the years (and from alot of other people here, on rgcc and ICC/ICS). My thing was REALLY WEAK when I started, and stayed that way for a good couple of years at least (partly through inactivity). It started with a blitz rating of maybe 1400 on ICC, and progress was pretty gradual. I had pretty much every bug imaginable including starting with a buggy version of alpha-beta! Plus tons of move generation bugs, all the usual stuff like castling bugs, promotion bugs, EP bugs, boundary bugs etc etc. My problems were caused by a number of factors: general lack of software engineering experience, a desire to do things differently, and of course a lack of chess programming knowlegde. Loved every minute of it though :-). Peter > >My original aim was not to write anything good. Juat anything. > >Frank
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.