Search found 1190 matches

by Zach Wegner
Thu Aug 09, 2007 6:49 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: tactical play or positional play for chess engine
Replies: 21
Views: 24702

Re: The Art of Evaluation (long)

This is best understood by an example: Assume that you have written a basic chess engine where the evaluation consists of material and piece square tables only. In this program, you add a single new evaluation term: A small bonus for rooks on open files. By doing this seemingly minor and innocent c...
by Zach Wegner
Sun Jul 22, 2007 7:20 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Monte Carlo chess
Replies: 5
Views: 3364

Re: Monte Carlo chess

This should probably go in the programming subforum. Anyways, I've experimented with it a bit, but in a different way. I replaced my qsearch with a loop that plays a certain number of moves for each side a certain number of times and averages out the eval after each sequence. The moves are also obta...
by Zach Wegner
Sun Jul 15, 2007 7:59 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Transposition Tables
Replies: 125
Views: 50439

Re: Transposition Tables

For castling you xor out the previous zobrist and xor in the new one, like this: board.hashkey ^= zobrist_castle[board.castle_rights]; board.castle_rights &= castle_rights_mask[from] & castle_rights_mask[to]; board.hashkey ^= zobrist_castle[board.castle_rights]; Then when there is an ep capture poss...
by Zach Wegner
Sun Jul 15, 2007 5:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Transposition Tables
Replies: 125
Views: 50439

Re: Transposition Tables

Simple. Just add a few more arrays. In my engine I have: HASHKEY zobrist_piece[2][6][64]; HASHKEY zobrist_castle[16]; HASHKEY zobrist_ep[64]; HASHKEY zobrist_stm; You could safely change zobrist_ep to 8 elements indexed by file, but I didn't bother. Castling is indexed by the 4-bit number discussed ...
by Zach Wegner
Tue Jul 03, 2007 6:18 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Iterative DTS
Replies: 17
Views: 10173

Re: Iterative DTS

Here's a simplified version of my search. The explicit gotos make the compiler create a jump table. The macros also do quite a lot to insure maximum readability. #define SEARCH_CALL(d, p, a, b, fm, dn, nt, rp) \ do { \ search_call(sb, d, p, a, b, fm, dn, nt, rp); \ sb++; \ goto ...
by Zach Wegner
Fri Jun 22, 2007 10:51 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: How much disk space on your development machine?
Replies: 13
Views: 6942

Re: How much disk space on your development machine?

How about music collections? Say it takes 8 MB for a typical 256 Kbps sampled music track. So that's 128 tracks per GB and about 128K tracks per TB. There are probably some collectors that actually have that much music. In terms of listening time, 1 TB of digital music would be very close to one ye...
by Zach Wegner
Tue May 22, 2007 2:49 am
Forum: Computer Chess Club: General Topics
Topic: Zappa on a 32 threads AMD (Barcelona core)
Replies: 12
Views: 6378

Re: Zappa on a 32 threads AMD (Barcelona core)

Looking at the CCRL data, it does not appear that Zappa scales better than Rybka: 1 Rybka 2.2 64-bit 4CPU 3105 +31 -30 73.0% -156.8 38.2% 369 78.4% Rybka 2.2 64-bit 2CPU 3052 +24 -24 74.3% -172.7 33.4% 665 65.3% Rybka 2.2 64-bit 2999 +35 -34 70.0% -128.8 47.9% 267 61.6% 2 Zap!Chess Zanzibar 64-bit ...
by Zach Wegner
Sun Apr 22, 2007 7:02 am
Forum: Computer Chess Club: General Topics
Topic: Why Christophe doesnt answer???
Replies: 25
Views: 13759

Re: Why Christophe doesnt answer???

I found the following link about DTS, Distributed Tree Search, and other approaches to parallel searching: http://www.cs.cmu.edu/afs/cs/project/jair/pub/volume9/cook98a-html/node3.html Just so you know, this is a different algorithm. DTS actually stands for Dynamic Tree Splitting. Here is the Bob H...
by Zach Wegner
Thu Mar 29, 2007 9:10 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: SEE on non-capture moves in main search
Replies: 62
Views: 72021

Re: SEE on non-capture moves in main search

I guess that the old PV move would disqualify itself anyway, as when you start the search at low depth its lower, but exact score would be satisfied from the hash table as it was already searched to higher depth. Really you would want to not probe the hash at that same node because it would be imme...