Author: Robert Hyatt
Date: 13:51:55 01/25/99
Go up one level in this thread
On January 25, 1999 at 16:22:11, James Robertson wrote: >On January 25, 1999 at 10:54:04, Albrecht Heeffer wrote: > >>On January 25, 1999 at 08:24:14, Fernando Villegas wrote: >> >>>Hi: >>>I have been amazed a little bit by the fact that in the long thread about bionic >>>as a clone or not nobody seems to have given data about how much Bionic is >>>really something new or not, to begin with. The only thing that has been said >>>about his presumed novelty is Dgeordge Vidanovic's statement that it is new; >>>only thing that has been said about his presumed clone quality is the afirmation >>>by Bob that a program that has changed only 1% of the code cannot be considered >>>something new. Well, which are the data to support one or the other statement? >> >>Hello all, >> >>It seems like a good time to introduce myself. I'm part of the Bionic team. >>I provided the hardware, worked through the test sets, added the end-game >>databases and did some optimisations. I was present at the Open Dutch during >>the first weekend. >> >>Let me first clarify some things: >> >>1) You want data? There is a complete website explaining a lot of details >>about Bionic. Please consult http://www.impakt.be/bionic/ There is >>also explained what was added and changes to the Crafty version >>we started from. >> >>2) All this was available before the start of the Open Dutch championship. >>We did not hide anything. Bionic Impakt is for a large part based on >>Crafty code, originally 9.26 and later based on 15.20. We never looked at >>Crafty 16.x as some people suggested. If you want to compare, you'd >>better use version 15.20 >> >>3) Hans Secelle is a good chess player. He added a lot of knowledge in >>the prescan routines. The effect is most dramatic on the positional >>problems of the Louget II test. Bionic scores 11 of the 14 problems, >>Crafty 15.20 only six! The results on all tests we have perfomed are >>on the website, including comparisions with Crafty 15.20 >> >>4) Bionic Impakt played on the same hardware using the same binary >>during the first weekend and the second weekend. The fact that Bionic >>scored better during the first weekend of the Open Dutch Championship >>is a conincidence. However is seems that a Crafty-derived Bionic that plays >>well is less acceptable by some than a Crafty-derived Bionic that score >>badly. Also, discussion starts again after Bionic wins from Crafty >>at the Winboard tournament. >> >>5) The claim by Robert Hyatt that Bionic plays the same moves than >>Crafty 16.1 is simply not true: >> >>>Vincent Diepeveen sent me the games from the first weekend of play. I picked >>>3 as that was all the time I had. I had crafty search each (on my quad P6/200 >>>which I figured was slower than the machine they used by a significant amount.) >>>I had crafty search for 10 minutes per move, and if it chose the same move >>>anywhere between 3 and 8 minutes as they did, I called this a match. I got >>>all but one move in those three games. Not a 'perfect' matching scheme, but >>>with the parallel search, it is non-trivial even on identical hardware. >> >>Several people tried this also, including Marcel Van Kervinck and >>Johan Havegheer. Marcel played through 513 moves and found that only >>406 matched with Crafty 16.1. This test can be verified by anyone who feels > >You shoot down this argument yourself in your previous paragraphs by saying "If >you want to compare, you'd better use version 15.20". These results now seem >meaningless. > >James > >>like doing so. We decided to post the complete loggings of the Open Dutch. >>You can download them from: http://www.impakt.be/bionic/odc98.htm >>You can inspect all setting, timings, moves, evaluations and so on. >>If you really want to do this in a scientific way: >>- do not use book moves >>- categorise chess-evading moves, 'obvious' moves, 'only' moves >>- do the same test also with some other program >> >>Good luck, >> >>Albrecht Heeffer Just so everyone knows, crafty v16.0 == crafty v15.20 + Eugene's new endgame database code (with a nasty enpassant bug in the probe code). 16.1 fixed the probe bug, but also included some eval changes.
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.