Author: Vasik Rajlich
Date: 03:25:41 04/06/04
Go up one level in this thread
On April 05, 2004 at 07:53:04, Bernd Nürnberger wrote: > >[...] >>>I put back killers now into my engine, and they indeed helps -- to less or >>>more account depending on the position (that's clear ...). >>>I will also try some experiments with the idea I read in another thread: >>>also take killer[ply-2n], killer[ply+2n]... > >>Interesting. Probably at most a very small win - but cheap anyway. > >After many experimenting I found out that in my engine, the best thing to >do is to use only *one* killer move; furthermore the ply-2n...ply+2n also >costs too much in Java (that bound checks on every array access!). > >[~ 2000 clocks/move] >> >>That is really fast. When Rybka was first running, it was also really fast. >>(Although not this fast.) At some point, you have to decide if you will keep it >>fast (ie like Fritz, Junior), or try some more expensive computation (ie. like >>Shredder, Hiarcs). > >I think you have to find a good balance between speed and knowledge. The >knowledge I have is clearly too less in the moment (pc-val, pc-sq-vals, >castling boni). > >[...] > >>Re. internal iterative deepening, it's much more important to order the first >>move well than to order the bottom half of the moves well. If you order your >>first move well, you'll fail high quicker, saving lots of nodes. If you get to >>the point of searching the bottom half of your moves, you are probably failing >>low anyway, and the order won't matter. > >That's surely try. I am doing hash (on fail: IID), than caps, three ordered >history moves, rest *unsorted* ... and that's very similiar to other engines. >Sorting all moves brings down nodes by only a 1-2% but punishing time by >5%-10% in my engine, so I decided it is not worth it. Btw - this will be worth revisiting as your engine evolves. If your engine gets generally slower, then the speed penalty will shrink. Also, if you decide to do a good fail soft, move ordering will help more. Cheers, Vas > >Greetings, Bernd
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.