Author: leonid
Date: 13:31:06 10/25/99
Go up one level in this thread
On October 25, 1999 at 11:11:00, Dann Corbit wrote: >On October 24, 1999 at 09:10:25, leonid wrote: > >>Hi! >> >>Would like to know how many Nodes Per Seconds can see usual game right now. I >>just found one message that spoke about this but message was partial. No >>computer speed or clear game name was done. >>Until now I could find this factor only on Rebel. I respect greatly this game >>for openness and performance. >Most of the programs will tell you a raw NPS number. However, I think they are >not comparable from program to program. The reason is that a slow searcher node >is not the same as a fast searcher node. HIARCS will get a very low NPS value, >and Crafty a very high one. But that does not indicate their relative strength. > I think the best indication of relative strength is the time to ply value (or >the number of plies at some time t). I agree with you 100%, but all the comparison business is very shaky. NPS is only useful as some highly imperfect aproximation to the final truth. For me the "final truth" is the "raw power" of the engin of the game. To obtain the "playing game" we actually add all kind of "happy tricks" to our initial "raw power". This permit us to overcome actual incapability of our game to see from given point up to the end of chess game. Leonid.
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.