Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Benchmarking chess algorithms

Author: Dave Gomboc

Date: 12:00:10 07/20/99

Go up one level in this thread


On July 20, 1999 at 14:17:21, Dann Corbit wrote:

>One way around these difficulties would be to create an object oriented model
>with clearly specified interfaces.  You can always put C programs in wrappers to
>achieve the same interface.  At any rate, with all the other variables the same,
>we are only testing one component at a time.
>
>When the tested component needs a service, it just requests it through one of
>the other objects.  Since the interface is spelled out, all we need is an
>initial functioning model to test all the parameters.  Once that is done, we can
>improve it a component at a time.  This will help the ability of the program,
>since any bottleneck will hog cycles and obsure the profiles.

Sounds like Dann has another ambitious project in line in case he gets bored
with C.A.P. :-)

Dave



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.