Thanks a lot, Patrick. Your tests are very useful to me, I sometimes refer to your ratings list before adding the engine to my division tournaments!
Also, May I add that it would be helpful if you also include the ratings of the previous version of Bikjump or other engines whenever you post the ratings of the latest one, so we could know how much the new version has improved relative to the older one.
Thanks again.
Last edited by swami on Wed Jun 11, 2008 7:12 am, edited 1 time in total.
Thanks a lot, Patrick. Your tests are very useful to me, I sometimes refer to your ratings list before adding the engine to my division tournaments!
Also, May I add that it would be helpful if you also include the previous version of Bikjump or other engines whenever you post the ratings of the latest one, so we could know how much the new version has improved relative to the older one.
Yes, this drop is unfortunately (although I could still hope for statistical noise of course ).
Version 1.7 is essentially 1.6.1, but with the added ability to query to Nalimov endgame tablebases (3,4,5 pieces) during the actual search, since I kindly received permission from Eugene Nalimov and Andrew Kadatch to "glue" their probing and decompression code into BikJump. This greatly improved endgame play, as BikJump has very little endgame knowledge by itself, and testing showed some very nice wins against stronger engines that do not have this ability.
But of course I also had to modify some other parts, and may have messed something up there. If you drop a piece in the middle game already, tablebases will not help you much.