Search found 466 matches

by Robert Pope
Tue Nov 06, 2007 12:34 am
Forum: Computer Chess Club: General Topics
Topic: Chessmaster 11 - first thoughts...
Replies: 13
Views: 6427

Re: Chessmaster 11 - first thoughts...

Chessmaster 10 added copy protection where it would not even install if you had virtual CD drive software on your computer, like daemon tools. Do you know if this is still the case? I'd love to upgrade from version 9, but not if it cripples other things.
by Robert Pope
Sat Nov 03, 2007 12:36 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Clear or not to clear the hash tables
Replies: 11
Views: 4790

Re: Clear or not to clear the hash tables

I thought of two things: 1. Are you validating the full hash key against the full key in the hashed position? You will get lots of false matches if you don't validate. 2. When Beaches played terrible moves, it was because of terminating in the middle of a search. The scores for the last root move se...
by Robert Pope
Sat Oct 13, 2007 3:42 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Tree dumping
Replies: 12
Views: 4608

Re: Tree dumping

I'm at least 80% there, with just a couple changes. I can't believe it made such a difference: 1. I discovered I had been limiting null-moves to ply 4 and deeper when my 3 ply search didn't have a single null-move cutoff. After looking at Crafty, I brought it down to ply 2. 2. After a fail-high, I w...
by Robert Pope
Thu Oct 11, 2007 12:29 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Tree dumping
Replies: 12
Views: 4608

Re: Tree dumping

Okay, I'm looking at a trace of WAC232 (iteration 3) and I'm trying to understand what I am seeing. Here's are two pieces of the output: 1 Qa4 d: 3.50 [ -0.43, 0.37] n:1066 SearchRoot(11) 2 Rxb8 d: 2.50 [ -0.37, 0.43] n:1067 Search(1) 3 null d: 1.50 [ 0.36, 0.37] n:1068 Search(0&...
by Robert Pope
Wed Oct 10, 2007 12:39 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Tree dumping
Replies: 12
Views: 4608

Re: Tree dumping

I expect the issue is in the search somewhere, like Uri said. I thought I had reasonable ordering rules, and I thought I had null-move working, but we'll see. Qsearch doesn't seem to be the culprit -- I turned it off and only saved 50%, which doesn't look too unreasonable, but that still leaves 2M n...
by Robert Pope
Tue Oct 09, 2007 1:30 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Tree dumping
Replies: 12
Views: 4608

Re: Tree dumping

I did the first comparison with Crafty19_20, but I enabled trace on Crafty20_14. Is it worth digging out older source code for a fairer comparison? Beaches is pretty vanilla still (nullmove r=2 plus some extensions).
by Robert Pope
Tue Oct 09, 2007 12:39 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Tree dumping
Replies: 12
Views: 4608

Tree dumping

I think the biggest thing holding back Beaches right now is the ridiculous number of nodes it is searching at each ply. I finally have an ideal candidate to dig into why my node counts are so terrible: WAC 232. Both Crafty and Beaches find the critical move at depth 6. The big difference is that Cra...
by Robert Pope
Tue Oct 09, 2007 12:16 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Compiling Crafty
Replies: 21
Views: 9357

Re: Compiling Crafty

I did. I haven't thought about it much recently, but here's what I think I did: Start new project import all .c files except dgt.c and dgtdrv.c import all .h files. In chess.h, I may have had to do a few defines. It looks like maybe I set: #undef UNIX #define NT_i386 I think once I had done that, I ...
by Robert Pope
Sat Sep 15, 2007 4:52 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Program for running test games
Replies: 2
Views: 1122

Program for running test games

I've always done my testing via 20-40 game matches against a single opponent directly through WinBoard. The recent discussions have given me the itch to enhance this process. What program is best to manage this? It seems like there are several possibilities -- JorisWBTM and Galis come to mind immedi...
by Robert Pope
Fri Jul 06, 2007 2:20 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Strange Perft And Divide
Replies: 4
Views: 2402

Re: Strange Perft And Divide

It's 123. After c7c5, there is an en passant capture possible.

I tested c7c5 on Crafty 20.14 and got the correct 30 moves, so I'm not sure where you are going wrong with Crafty. Are you actually making the move c7c5, or are you pasting the resulting fen into the programs?