Computer Chess Club Archives


Search

Terms

Messages

Subject: Rybka bugs comment

Author: Vasik Rajlich

Date: 02:39:08 12/08/05


Hello,

I appreciate the great feedback from the testers. Rybka played over 5,000 games
in my testing, but under very limited conditions, and many new things have
popped up after the Beta release. Here is a list of major known problems with
comments:

1) Mating problems - essentially fixed (in full release). Rybka still won't
always play the very fastest mate but will be _much_ more efficient. A really
full solution requires even hash table changes and will be available as a patch
to the first release.

2) Displaying mate lengths - will be available as a patch to the first release.

3) Displaying mate variations in analysis mode - fixed.

4) Displaying multiple variations in analysis mode - fixed.

5) 50 move repetition problems - fixed.

6) Support for full UCI protocol - fixed.

7) Underpromotions - will be fixed as a post-release patch.

8) Crashing & hanging problems - some fixed, some being fixed. Obviously the
highest priority at this point.

9) EGTB support - added.

10) Endgame knowledge - this is for Rybka 2.0. It should not be hard in
principle, in fact I expect it to be easier than middlegame knowledge. There
just wasn't time. I do apologize to the endgame lovers out there :)

11) NPS - please don't worry about this number. Rybka uses the same CPU cycles
as everyone else. Search is a very weird and complicated thing, every engine is
searching 99.9% nonsense anyway.

Again, I appreciate all of the great feedback and look forward to being ready
with a stable 1.0 in a few days.

Best regards,
Vas



This page took 0.02 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.