Author: Uri Blass
Date: 06:27:12 12/09/02
Go up one level in this thread
On December 09, 2002 at 08:57:04, Bob Durrett wrote: >On December 09, 2002 at 02:45:43, Sandro Necchi wrote: > >>On December 08, 2002 at 05:12:47, Uri Blass wrote: >> >>>On December 08, 2002 at 03:07:30, Helmut Conrady wrote: >>> >>>>On December 08, 2002 at 02:04:35, Sandro Necchi wrote: >>>> >>>>>On December 07, 2002 at 22:19:13, Tim Blankenship wrote: >>>>> >>>>>>havent heard but 1 or 2 things about it..is it suppose to be strong or does >>>>>>anyone know...thx >>>>> >>>>>My best guess for the new Shredder 7 is about 50 points stronger. I wrote this >>>>>in CCC. >>>>> >>>>>Talking with other testers I got this confirm by them, too, >>>> >>>>Yes, thats right, Sandro: >>>> >>>>E.g.:I have done a test match Fritz 8-Shredder 7 >>>> >>>>conditions: >>>> >>>>Fritz 8_GUI >>>>Athlon 1700+ >>>>50 games >>>>no TBs >>>>10´+4`´ >>>>Powerbooks 2002 (varietety of play: max; minimum of games: 10). >>>> >>>>29: 21 for Shredder (+23 -15 =12), = 58%. >>>> >>>>The other results are very impressive, too. After some hundred games Shredder 7 >>>>is within the 95%-margin stronger than Deep Fritz 7 >>>> >>>> >>>>I predict Shredder will top the next SSDF-list. >>>> >>>>Helmut >>> >>>I am interested to know the results of shredder in tactical test suites like the >>>GCP test suite. >> >>Hi Uri, >>> >>>Is shredder better than other commercial programs in tactics? >> >>I have not made specific test/comparison, but only looked at the played games. >>What I can say is that Stefan did a huge work on the engine in all fields like >>speed up, chess knowledge add, new settings and new search. Also the selective >>search was increased. >>My impression is that it is tactically faster than 6.0 a lot and that loves >>complicated tactical positions. >>However the most improvements are on positional understanding combined with >>tactics. > >I do not doubt that it is true that the program has positional understanding. > >But, to me, the interesting question is: HOW does one determine &/or verify >that a program has positional understanding? What is the method for measuring >this understanding and how to come up with a measure? > >Bob D. Here is an idea that I thought about. 1)play matches between 2 programs and care to use hardware difference such that the result is going to be close to 50%. 2)Look only at the games when both programs believe that their position is better based on evaluation for many moves and the difference in evaluation is at least 0.5 pawns for at least 4 moves so it is probably not about tactics because I do not expect one to outsearch the other by more than 8 plies. The program that win most of the games in 2 has probably better positional understanding. 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.