Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Puzzled about testsuites

Author: Tord Romstad

Date: 11:24:16 03/09/04

Go up one level in this thread


On March 09, 2004 at 12:02:47, Michel Langeveld wrote:

>Hi Tord!

Hi Michel!

>Thanks for answering
>
>I am affraid you are right ... I test most of the time a lot and make tiny
>steps. What timecontrol do you use for testing games?

I am not a very patient person.  In most cases, I play 2 minutes/game
or 4 minutes/game, with a 1 or 2 second increment.  Occasionally, I play
at ridiculously fast time controls like 15 seconds/game.  The only point
of playing these ultra-fast games is that I hope to eventually port my
engine to slow hand-held devices.

>>The complete ecm.epd would be no better.  All positions are tactical.
>>LCT2 is great, but unfortunately I don't know any other suites of
>>similar type and quality.
>
>What timecontrol do you normally use to test this? Do you look also for each
>position to number of nodes searched. ply depth solution found?

Are you asking specifically about LCT2 here, or about test suites in
general?  LCT2 comes with instructions about how to run it.  You run
each position for 10 minutes, and only solutions which is found and
kept until the 10 minutes have passed are counted.

For other test suites, I don't run them often enough to have any fixed
routine about how to run them.  The only one I use regularly is WAC,
which I use mainly for debugging purposes.  I let the engine think
5 seconds/position, and simply check the number of solutions found
when it is finished.  If less than 295 positions are solved, I take
a closer look at the unsolved positions in order to find out if I have
introduced some new search bugs.  My latest version solves 298/300 at
5 seconds.

Tord



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.