Author: Vincent Diepeveen
Date: 08:48:06 12/13/05
Go up one level in this thread
On December 13, 2005 at 02:16:50, Chrilly Donninger wrote: >I experimented recently with a Shredder-style search in Hydra. The >single-processor Shredder/Hydra completly demolished Shredder. If two programs >are similar, the strength difference is enlarged. Its therefore a bad idea to >tune a programme against itself. >But the Shredder/Hydra made only 40% against Rybka. Changing back to the >standard Hydra-search its between 75-80%. Rybka is regularily "killed" in >king-attacks. As noted before, this numbers are for Hydra-single-processor. The So that's 1 pc processor and 2 fpga boards? >PC-programm is running on a 3.2 MHz Pentium 4. Time control is 30secs/move. A >standard-opening set similar to the Nunn-openings is used. > >Changing the search is not only a tactical matter. The playing style is to a >large extend also influenced by the search. If two moves are from the evaluation >point similar, the programm usually plays the one with the larger search tree. the program plays the move that gives the biggest score. If you are winning that's the line with the larger search tree, if you are in trouble or when the program is too happy in root about its position, then that's the line with the shortest search tree. >Or in other words: The lines which are extended. The Shredder/Hydra played >over-aggressive, whereas the classical Hydra with the right dose. >One conclusion of my experiment is: Rybka seems to be fairly tuned against >Shredder. This is always the fate of the leader of the gang. In the future other >programs will be tuned against Rybka and it will be much more difficult to stay >on the top. I feel it is wrong to conclude rybka to be tuned just against shredder. It's better to conclude that the way to beat rybka is by good positional play, the shredder search simply doesn't give that, as it has a worst case performance, and the normal hydra search has no real worst case. When you get tactical outsearched by Rybka, it's important to not show a worst case. >The experiment shows also, that it is fairly easy to tune against one programm. >The problem is to find a solution which works against all. >Chrilly It is not so easy easy to tune against just 1 program, but it's wrong to conclude IMHO that everyone is tuned against shredder. When you test objectively at home, you will see Rybka scores better against Fritz, than it does against Shredder. Simply because Rybka is everywhere better than Fritz. At least there is 1 thing shredder has which fritz doesn't have, and that's that shredder is outsearching Rybka in terms of iteration depth shown (yes, i know it's thin plies). In tournaments Shredder shows 20 ply at a quad machine, Rybka won't. Fritz however gets outsearched everywhere. Online everyone is playing only with shredder9.ctg instead of his classic book, and online a few other things go wrong, such as random moves and such. that confuses the audience with objectivity. Tests at home are more interesting. It's better to conclude that Rybka has no real worst case behaviour like Shredder has. And if you eliminate your worst case, when you are a favourite to win matches, despite that rybka tactical outsearches hydra everywhere, then that's always good. Rybka is excellent if a program has a worst case. Every game one loses against it, when i analyze it, there was something wrong in that game which a 1900 rated player already spots, but which for some reason came onto the board. Please add to this that another reason why your shredder style wins from it less than normal hydra is perhaps because hydra's king attacks are already based upon too little knowledge. It just goes for it without having any doubts. Against a tactical stronger opponent such attacks don't work then of course. This whereas you stated above that normal hydra is less agressive. Vincent
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.