Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Engine comparison solving test suite - Rybka on top

Author: Paul Jacobean Sacral

Date: 04:09:46 12/06/05

Go up one level in this thread


On December 05, 2005 at 17:40:50, John Rose wrote:

>On December 05, 2005 at 17:11:50, Paul Jacobean Sacral wrote:
>
>>Did you set max. plies = 99?
>>
>>It's recommendable, or required actually, for testing with time limit.

>I dont see a max ply option. I do see an option called "extra ply" which
>defaults to 1

Correct. If you don't set this to 99, you can get wrong solutions. For instance,
a correct move after 0.5 seconds and 3 or 4 plies only, where actually the
engine would switch to a wrong move at ply 5 and need more time, or even don't
solve at all. If you set this to 99 extra plies, it will search each position up
to the time limit (not up to 99 plies :-)) making sure that it is a "true"
solution and not just some more or less random noise, from very low ply depths.

This is only a methodical remark and doesn't mean that I doubt the good Rybka
results. I have already seen some very good test suite results reported, today.

Yours truly Paul J. Sacral



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.