Author: Ferdinand S. Mosca
Date: 10:21:45 08/08/02
Go up one level in this thread
On August 08, 2002 at 08:44:28, Marc van Hal wrote: >On August 07, 2002 at 18:34:16, Ferdinand S. Mosca wrote: > >>I have tested the new Aristarch 4.4 of Stefan Zizproth against the world >>champion Junior 7 at 2hr/40moves+1hr/20moves+30min. Aristarch seemed to have >>the best time allocation based from a position. Like thinking long in an unclear >>position or something that needs time to get an upper hand. Easy moves are too >>easy too for this engine, it has a relatively high searched depth in the middle >>than in the endgame (maybe i got lost on the parameters, i did not use the >>default one), Junior has a high searched depth but Aristarch is right behind >>there. This engine likes to attack the king. Here's some positions with unusual >>double digit time allocation in minutes. > >I gues you need fast hardware for having it play good I think just more time is needed, to me I want to see an engine releases its all knowledge and search techniques as much as possible, since time is the problem, thinking long at the right time is needed. I still have to closely look at the updated text file where parameters are commented. I am interested on the time allocation of this engine. This engine is fast too and I thought that it can afford to calculate various attacking schemes on opponent's king. >Maybe a program which indeed takes advantage of it >I played a match between Anmon and it and te results where not so superior Yes because Anmon is also a strong program. >it also doesn't use the full hastables it gets when playing 30sec/movegames > The games it won where actualy out mistake from Anmon >It has trouble to see moves which lockup a piece maby setting pins diferently >might help. My observation here is something about the development, it has the tendency to attack opponent king prematurely. For weak opponents this is ok but against equally or slightly superior opposition this premature attack won't work, but this is not a problem at all, time is needed to tune it for optimal settings. Dinan >The score I had was in favour for Anmon though >Strangely enough also >Which actualy is strange Aristarch had trouble in finding some pins too >in some cases if it could choose to pin a piece or not it choose for the latter >Which made positions which it should win to a draw >The program also did not have all options in the Shredder5 interface >as mentioned in the readme file. >In the Arena Interface options did not even become active >But this are still only quick conclusions. >Perhaps i find some more conclusions later. > >Regards Marc van Hal
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.