Author: Joel
Date: 17:49:26 12/23/02
Go up one level in this thread
Hey Martin, My engine is not complete yet, but I am having (what I think is) acceptable results with this method: I first generate the moves in the order: capture promotions, promotions, captures (I need to work on ordering these somehow, at the moment it is a bit simple), castles, and the rest of the moves follow. Before I use this list, I see if my transposition table has a suggested move, and I try that. If it doesn't and the depth to which I am searching to is high enough, I call my search function again at a reduced depth to get a suggested move - currently D-2 since due to horizon effect (I have no quiescence search yet) my evaluation oscillates heavily. The reduced depth I use needs tweaking for sure. Not sure of the effectiveness of my move ordering methods. I seem to have a reasonably low branching factor, but I need to look into this much more. Also, can anyone give me their opinion as to whether 9 plies from the starting position in 15 seconds is good, bad or mediocre? Keeping in mind I am using just AlphaBeta with MTD(f) with absolutely no pruning beyond what AlphaBeta gives you. My system is a Athlon XP 2100, and my NPS is around 500K and I have a 32Meg hash table. Oh, and my evaluation is quite small at the moment, 400 lines approx. Regards, Joel
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.