Author: Robin Smith
Date: 16:10:51 01/27/99
Go up one level in this thread
On January 27, 1999 at 18:50:31, John Coffey wrote: >I saw a figure of Fritz 5 searching about 200,000 NPS on a 200 MHZ MMX Pentium. >Crafty claimed about 80,000 nodes on the same hardware. > >However, that doesn't tell you much really. When you think about it, you could >write a program that searches 1M nodes per second. Current programs would play >better than it would because they use techniques to reduce the branching factor >and thus reduce the size of the tree. Those techniques take many clock cycles >which reduce the NPS, but they save in the long run because they don't have to >search as many nodes. > >These techniques to reduce branching allows Crafty to only increase the number >of nodes by a factor of 3 with each increase in ply. (This amazes me, but >apparently it is true.) > >John Coffey Yes, yes, what you say is true. But what about looking at nodes per second on programs that have approximately the same ratings on the sweedish ratings list. It still seems it would tell something about the relative performance change one might expect.
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.