Author: Daniel Shawul
Date: 04:10:29 04/08/04
Hi
Here are some questions
1.In case of check, Do you think generating only legal moves will be faster
thank generating and then checking for legality? I need to count the no of legal
moves for extension purposes.
2.In Ed's paper there are some tricks done at horizon.
trick one if(score - (margin + largest hanging piece) > beta)
don't do quiescence
trick two if(score - 900 > beta)
don't do quiescence
My question is every body does
score=eval()
if(score > beta)
don't do quiescence;
So i don't see how the tricks work?
3.I have implemented attack tables
2 bits for pawns
2 for bishop/knight
2 for rooks
1 for queen
1 for king
The reason that i chose this one over ed's is
1.if i need to change to incremental updating this one will be more
comfortable
2.can tell exact nos of each attacking piece [easier for SEE]
3.incase of overflow,for example 2 queens will promote to 1 rook
whick "I think" is good for SEE. It is very rare the data gets corrupted
by lots of attackers.
The only drawback i can see of this method is if there is something that you do
with only the type of attackers[ie you don't need counts] you need to use
a 256 array than a 32. What do you think?
3.I thought i could get In_Check and Checks for free if I use attack tables.
But there is trouble. If i need to know whether a move checks , I need to
generate the attack tables which is very costy. Same applies for In_Check.
For example I need to know whether a move checks if i am using futility pruning.
So i am doing an incrementally updated checks and in_check to facilitate this.
Infact it is doing very fine. The in_checks require we are not in check before.
So I need to generate completely legal move when i am in check.
4.Is history heuristic that important? I know it gets corrupted at larger
depths. scoring non capture moves by their piece square values seems to work
better for me than history.
5.How do i measre my move ordering performance?Is it by counting only fail highs
of the first moves searched. It seems to me doing this measures performance of
hashtable since the hash move is the first searched. Or is it counting fail
highs of first x moves??
if what i wrote made sense to you,and have some ideas please post back.
Thanks
Daniel
This page took 0.01 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.