Author: Dan Newman
Date: 14:18:26 05/15/98
Go up one level in this thread
On May 15, 1998 at 02:10:11, Bruce Moreland wrote: > >On May 15, 1998 at 00:31:19, David Blackman wrote: > >>However floating point add, subtract and compare will be slower than >>integer add, subtract and compare on just about anything. The inside >>loops of chess programs tend to do a lot of add, subtract and compare, >>but very little multiply and divide. So integer arithmetic is probably >>the best choice for chess. It would be different for graphics. > >Or even *no* multiplies and divides. > >bruce I avoid integer multiplies like the plague, however I do use them in one place: updating the history table. I stole this idea from Crafty. I just add the square of the depth remaining (+= depth * depth) to the table entry. This keeps the entries from blowing up on truly deep searches. I had been using += (1 << depth), which I think I got from Schaeffer's paper. IIRC, when the depth gets to about 30 or 31 the 32-bit table entries start getting messed up. -Dan
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.