Author: Uri Blass
Date: 11:51:44 11/27/00
Go up one level in this thread
On November 27, 2000 at 13:55:13, Frank Quisinsky wrote: >On November 27, 2000 at 13:48:18, Uri Blass wrote: > >>On November 27, 2000 at 12:02:35, Frank Quisinsky wrote: >> >>>Hi, >>> >>>here can you find information about the Gandalf 4.32g beta test. >>>We think a good idea for user which have interest: >>> >>>CCE forum: >>>http://f22.parsimony.net/forum41668/ >>> >>>Do you have questions about Gandalf 4.32g ? >>>(ultimativ TOP chess program, is running under a lot of or better all new >>>commercial GUIs (playing software) as a "WinBoard Engine") ! >>> >>>Furthermore, beta testers copy results and error messages (important for Steen) >>>in this fora. >>> >>>This is a service from Gambit-Soft and Steen Suurballe. >>> >>>Thanks and ... Have a nice day ! >>> >>>Best >>>Frank >> >>I respond here to your message because of the problem of writing in that forum. >>You asked it to do longer time control games. > >Yes, sorry ! >I see the problem in the Parsimony Forum :-( >I hope not a long time ! > >> >>I know that Gandalf is better at longer time control. >> >>Gandlaf4.32f did better results in my 15 minutes game against Fritz6light >>relative to my 5 minutes games. >> >>The reason that I did 15 minutes games is that I wanted to compare 4.32f with >>4.32g and I wanted to use the same conditions as 4.32f. > >Yes a good idea ! > >>Using longer time control will not give me information about comparison between >>4.32f and 4.32g because I almost did not do games at longer time control with >>4.32f. > >OK, I don`t know this ! > >>I can add that gandalf 4.32f liked more the a3 f6,a3 f5,a3 g6,a3 g5,a3 h6, a3 >>h5,a3 Na6, a3 Nh6, a3 Nf6, a3 Nh6 opening and got better results in these games >>and I tested only the a3 a6,...a3 e5 opening with 4.32g so I expect 4.32g to get >>better result if I continue the match of 15 minutes per game. > >>I will use 64 mbytes hash for the next games because you say it does better >>results with 64 mbytes. >>I did not decide if to use longer time control or if to cntinue the comparison >>between 4.32f and 4.32g > >I have not a direct comparison but when I see from different user results with >64 or 128 MB are the results with 64 MB better. I am not sure why but it is >interesting. > >Thanks ! I started a test with 64 mbytes for gandalf from the same position at the same time control. One of my ideas is that maybe 128 mbytes help the computer to be slower. I think that a good idea for chessbase is to give the number of nodes per second in the pgn file because in that way it will be easy to find if the computer is slowed down by a significant factor. I found in the past in chessfun's games that one computer was slowed down by some factor and I suspect that it happened in my games and I have no idea how to discover it by only looking at the pgn. If I get the number of nodes per second then I can see if there is a significant change in the speed only by looking at the pgn. 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.