Author: Robert Hyatt
Date: 19:03:22 10/29/98
Go up one level in this thread
On October 29, 1998 at 15:29:33, William H Rogers wrote: >I just ran a compiled version of my program. I chose a 4 ply game, and I do not >do quesience or any thing else. It's blind behind 4 plys. >There are roughly 487,780 possible positions it I start with e2e4 for white. >It only takes my program 7 seconds to get the answer e7e5 using minimax. >Does this mean that I am searching 69,683 nodes per second, and if so is this >good compared to other programs. >While reading posts on the board, I noticed that it is faster to generate all >moves first, and check for legal in the actual minimax play of the moves. >I think that I can do this and save 487,780 incheck calls. >My program generates a move and then goes to see if the king is check, if so it >deletes that move from the list, and as there are not that many checks, at least >in the beginning of the game, doing a check later would save tons of time. >Can someone answer my question. >Thanks >Bill Rogers that's not a bad speed. but without alpha/beta it is impossible to compare NPS with other programs. IE for minimax you don't do any wasted work since you must search every node. So time spent sorting moves is wasted (minimax doesn't need *any* move ordering of any kind) while generating all moves at once is good since they all have to be searched... Get alpha/beta going next. and you go twice as deep in the same time...
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.