Author: Robert Hyatt
Date: 18:46:21 07/24/00
Go up one level in this thread
On July 24, 2000 at 19:50:54, John Coffey wrote: >On July 24, 2000 at 09:14:43, Robert Hyatt wrote: > >>On July 24, 2000 at 07:29:03, Larry Griffiths wrote: >> >>>On July 23, 2000 at 21:59:58, Robert Hyatt wrote: >>> >>>> >>>>This is on my list to fix, as it is not quite easy to search > 2B nodes and >>>>wrap the counter around... >>> >>>I understand Robert :) >>> >>>Looks like I would have to combine my captures and moves into 1 bitboard to make >>>my bruteforce faster if I want to compete with Crafty's perft command. >>>Im sure that adding castleing and pawn promotion will slow my program down even >>>more. >>> >>>Larry. >> >>There was a typo in my response. "not quite easy" should be "now quite easy" >>as searching over 1M nps will wrap the counter inside an hour. >> >>Castling and EP really shouldn't slow you down, as you will skip that code >>most of the time. Just make the "should I do this?" test as efficient as >>possible so that it doesn't impact you when you can't castle, which is most of >>the game... > >Please tell me what hardware allows you to search 1M nps? > >John Coffey My quad xeon/550 will do it. A 21264 (single cpu) will also do the trick. (Alpha cpu)
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.