Author: Enrique Irazoqui
Date: 04:04:36 12/17/00
Go up one level in this thread
>I am interested to know which program is best in this secret test Results of my test are currently a royal mess. I got faster machines and I must run the test again with all the programs. It takes time and right now I feel quite lazy about it. By the way, programs that use 2 processors, like Crafty and Deep Fritz, don't to give very consistent time/solution answers to positions. Why, oh why. :) >What are the results of the following programs: > >1)Deep Fritz(Does it have better result than Fritz6a) On a dual machine DF does much better, of course. On a single, roughly the same. I was told that the main improvements in DF over F6 are not related to search but to king safety and the handling of closed positions, so DF doesn't have to do better in a tactical test. >2)Chessmaster8000 I don't have it yet. >3)Tiger13 A tiny bit better than Tiger 12. >4)Gambittiger Not yet tested. >5)Century3 A tiny bit better than Century 1.2. >6)Gandalf4.32g Similar to Fritz 5.32, or 75 points less than F6a, but I expect it to show the J6a syndrome, that is doing better in games than in tactical tests. >I am interested to know the number of correct solutions at different time limit >to get an impression which program earn most from time(the program that earns >more solutions when the time limit is increased is probably the program that >earns most from time) There are different ways to look for this, but first I have to run the test again on the same new platform. Anyway, until one year ago my test gave results that were practically identical to the SSDF list. Then Shredder 4 and even more Junior 6 did quite a bit better in real games than in my test. This year I see that clearly improved programs, like Tiger 13 and Century 3, don't perform significantly better in my test than their previous versions. Since my test is exclusively tactical, this must mean that improvements took place in other areas. This is a bit puzzling, given the fact that according to Ed Century 3 has a much better search than C1.2, but C3 does only marginally better in my test. I would expect the improved search of C3 showing much more in a tactical test, but no. Again, why oh why. :) Just for fun: based on the games I played and also on my test, I expect that programs will be rated in comp-comp roughly like this, picking F6a as reference: DF +20 F6a, Tiger 13, Gambit Tiger, 0 Shredder 4.22 -30 (S5 is supposed to be better than that. Equal to F6a? I don't have it yet, but a copy is hopefully on its way.) J6a, Gandalf 4.32g, -30/-40 C3, -70 N732, -80 J5, H732, Crafty 17.14, SOS, -100 Enrique
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.