Author: Sven Reichard
Date: 14:08:08 09/10/02
Go up one level in this thread
On September 10, 2002 at 15:02:51, Jean Bouchat wrote: >Hello, >we are writing a new chess-engine (not ready for release: it can only move pawns >and knights) and I wonder how fast is your move generator functions. Let's take >for example the starting position. I called ours 1 million times on a PIII 733 >MHz. This includes the legality of the moves(no check), memory allocation for >the 20 new positions and freeing of this memory at each call. It takes 16 >seconds. >Is it fast? >Jean Dear Jean, IMHO it is a bit difficult to compare the speed of move generation among different programs. The problem is that you might include things in the generation (like check detection) that others would do only when the move is actually made, or even later. What I do to measure the "raw speed" is generating the (pseudo-legal) moves, making them, test for check, and take them back. Making moves involves updating certain data structures like hashtables. In the initial position I reach about 300 kHz on a 1 GHz Athlon (~3000 cycles per move). Bear in mind though that the evaluation function and even the search algorithm use much more time, bringing the total consumption to about 20k cycles/node. That being said, your 1.25 Mhz for move generation look reasonable to me. Sven.
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.