Computer Chess Club Archives


Search

Terms

Messages

Subject: The secret of a small tree

Author: Scott Gasch

Date: 10:17:45 10/09/99


Hey,

My branching factor is too high, or something.  I have eliminated my explosive
qsearch but my main search tree is much larger than other programs'.  The only
answer I have heard for this is "your move order is messed up."

My move order is as follows: 1) hash move, 2) winning captures ordered by SEE,
3) even captures (determined by SEE), killer1, killer2, rest ordered on history
heuristic.  Am I missing something?

I update killers and history at a fail high (beta cutoff) and a PV node.  I do
not do anything with a root move list and I do not do internal iterative
deepening (don't understand it yet).  I am doing a PVS and in my tree dumps I
see I am having to research sometimes, esp in low depth searches.

As an example of my branching problem, position fine 70
(8/k7/3p4/p2P1p2/P2P1P2/8/8/K7 w) at 20 ply I am over 1 million nodes whereas
every good program is under 100k.  Likewise after 1. a2a3 at seven ply I am at
70k while others are more like 30k.

Thank you once again for the help,
Scott



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.