Author: Dan Honeycutt
Date: 15:58:40 05/23/04
Go up one level in this thread
On May 23, 2004 at 11:49:59, Andrew Wagner wrote: >Yes, finally I'm posting about something other than the authors-only tournament! >:) > >Joel (author of Bodo) and I have been working on putting together a move >generator test suite. I posted about this once or twice before, but all the >kinks are now worked out! The idea was three-fold: > >1.) Make the test suite hard enough that your engine would be very unlikely to >pass unless the move generator works. As Russ Reagan pointed out before, I can't >guarantee that your move generator is perfect if you pass this test, but I can >come pretty close :) > Hi Andrew: Good job, and timely - I was about to ask you what had become of this project. I ran your positions and matched your numbers. Then I introduced a bug (actually unfixed a subtle bug I used to have). Reran the positions and matched the numbers again - thus proving what Russell said. Here is a suggested addition to your positions: [D]8/5p2/8/r3P2K/k2p3R/8/2P5/8 w - - 0 1 The problem Bruja (which does legal move generation) had with this position was after c4 it didn't realize that capturing enpassant is illegal. The black pawn on d4 did not "appear" to be pinned as the white pawn on c4 provides shelter. For this position I get (without the bug): ;D1 12 ;D2 138 ;D3 1961 ;D4 28763 ;D5 463193 Thanks to you and Joel for the efforts. Dan H.
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.