Author: James Swafford
Date: 19:30:41 11/15/00
Go up one level in this thread
On November 15, 2000 at 21:12:48, Michael Neish wrote: > >Okay, now here's something I just don't understand. > >On my slow 240 MHz PowerBook, the freeware programs MacChess and Sigma Chess 5.1 >reach 120-150,000 nps in the middlegame. MacCrafty 17.12 does around 20-30,000 >nps. My own program, inferior to all three, does about 20-25,000 nps. > >Why such a great difference in speed? I don't even see how taking knowledge out >would manage it (not that I want to do that). Just for a laugh I tried material >only evaluation, and got only about 45,000 nps. It seems impossible to me even >to reach 80,000 nps, let alone higher. > >Can anyone please explain the vast difference in speed between all these >programs? I trust that my program is not wasting a vast amount of its time, >although it's surely not optimised. > >Cheers, > >Mike. First, let's compare apples to apples. How do you count nodes? Are you counting frontier nodes, or just interior ones? By optimized code.... do you mean compiler optimized or hand optimized? Which compiler are you using? Is your position structure large? If so, are you doing a make/unmake or copying the position? What about your search.... are you using a quiescence search? What about lazy evals? What kind of move generator do you use? Does it separate caps/noncaps or are they all gen'd together? Do you ever fail high before generating moves? Are you using SEE or mva/lva? There are lots of variables here. :-) -- James
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.