Computer Chess Club Archives


Search

Terms

Messages

Subject: One and half month old mystery

Author: leonid

Date: 13:50:25 06/13/99


How game count number of positions that it see each second?

As simple as this question can sound, it is tricky one. Problem rely
inside of ply zero. In the ply much higher (that lowest ply) number of positions
seen is the number of moves deposed on the board for evaluation.
But on the ply zero you don't need to do this at all! On the lowest ply
you can have around 4 moves [nodes] to evaluate but you do this without deposing
even single one on the board. Everything is done while
generating those three or four moves. When I tryed to find how many
positions in one second my game can see, I was all the time confused.
When not counting the 3 or 4 moves generated on the level zero [because no
official deposition of moves is done]I recieve very mederate, if not very
mediocre number. It was around 40 - 45k per second as a "big shot". When
I see this I am ready to drop my game because my chances to make it good
are simply not there. But when I artifically obligate my game to
see on the lowest ply all the moves, like everywhere (by deposing them
on the board for studying) my numbers goes up and I feel me happy.
On my Pentium 166 it can go as far as 250k per second, but generaly around
150K. With the last numbers, I can have some confidence to end well my
game. Never mind if some extra ticks will be eventually lost when the
game will become more complicated in the future. I will just arrrive to
the normal speed at the end of programming.

Now You see why I want finally find what is the reality behind the
positions/second numbers. How calculation is done usually inside of
the chess game? Solve a mystery for me!!!

Leonid.



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.