Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Benchmarking chess algorithms

Author: Dann Corbit

Date: 10:09:11 07/20/99

Go up one level in this thread


On July 20, 1999 at 12:53:31, KarinsDad wrote:
[snip]
>What if your program only generates legal moves and there is no unmake?
>
>Again, I am skeptical. All of this is similar to the nodes per second issue
>where each implementation (sometimes from one version of the same program to the
>next) is different.
If we are just talking about move generation, then NPS is a valid mark.  Some
programs put a ton of energy into being selective in their search, and the NPS
goes way down.  In such a case, what we really need to measure is plies.
However, the NPS for the move generator itself is still very valuable.  A slow
searcher that has a slow move generator will benefit if it is replaced by a fast
move generator.  If intelligence is bolted right into the move generator, then
bolt it onto a fast move generator instead of a slow one.




This page took 0.01 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.