Author: blass uri
Date: 05:17:40 04/27/00
Go up one level in this thread
On April 27, 2000 at 07:01:06, Mogens Larsen wrote: >On April 27, 2000 at 05:10:16, blass uri wrote: > >>On April 27, 2000 at 02:16:48, Chessfun wrote: >> >>> >>>All games on one Cel 433. >>>Ponder=off. >>>Tablebases used at 25 and longer. >>>Nunn 1 positions. >>>No opening books are loaded. >>> >>> 1 min game Fritz 6a 14.5 - 5.5 Crafty 17-10 >>> 2 min game Fritz 6a 14.5 - 5.5 Crafty 17-10 >>> 3 min game Fritz 6a 13.0 - 7.0 Crafty 17-10 >>> 5 min game Fritz 6a 15.5 - 4.5 Crafty 17-10 >>>10 min game Fritz 6a 13.0 - 7.0 Crafty 17-10 >>>25 min game Fritz 6a 13.0 - 7.0 Crafty 17-10 >>>60 min game Fritz 6a 15.5 - 4.5 Crafty 17-10 >>>Tourney times Fritz 6a 14.5 - 5.5 Crafty 17-10 >>> >>>Once I figure out how to autoplay from the Nunn 1 opening >>>positions I will play all games with ponder=on. >>> >>>Thanks. >> >> >>Total result 113.5-46.5 indicates 167.5 elo difference between Crafty and >>Fritz6a >>when there is no pondering in the nunn game. >> >>tourney game indicates 180 elo difference and there is no significant >>difference. >> >>The reason may be that nunn1 positions are not good and it is better to use the >>nunn match with positions for practical games(It seemed to me that crafty earns >>more from time based on the ssdf results but maybe I am wrong and Crafty is not >>more than 100 elo weaker than Fritz6a also in blitz). > >I think I know by now, why the Nunn position(s) has been used in this test and >the reason is defendable. But in general I don't think it's a good idea to use >them for testing purposes. A computer chess program is a package deal. If you >purposely remove certain elements, ie. opening book or ponder, you get results >that are purely of academic interest. I your're adament of making a direct >engine to engine comparison _everything_ should be removed including EGTB (if >you're real zealous a third party GUI should be used). I disagree here. the possibility to use the nalimov tablebases is part of the engine. This is not the same as opening book. I am interested in the ability of programs to analyze positions and nalimov tablebases are totally relevant because programs can use them not only at the root. If using tablebases of opening and middle game position can help programs not only at the root then I think that these tablebases should also be allowed. I think that practically there is no perfect informatio about opening positions when programs are out of book and using the idea of big tablebases of opening positions and evaluation can be a problem because of the fact that the programs will be slower(mainly in cases when there are no tablebases hits and you waste time about every node) This is not the case in the nalimov tablebases because in this case you do not need to waste time after a move to check for tablebases hits if it is not a capture in the endgame And no starting position >should be chosen, which includes Nunn. That's my opinion. How do you want to test program with no starting position. It is impossible to play games with no starting positions. I think that using starting positions like the ssdf positions out of book is better because these are positions from practical games and the book of crafty in the ssdf games is similiar to the book of Fritz6a. Crafty find itself in both sides of these positions in the ssdf games so these positions are relevant to test crafty. There is only one problem that it is possible that crafty learned in the ssdf games not to play opening that Fritz plays and the opposite so the results may be different relative to the ssdf results but I think that humans are interested in programs mainly for analysis so my idea of using the ssdf positions out of book is clearly relevant for deciding which program to use for analysis even if it may give wrong results about rating. Uri
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.