Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What's the best tactical engine?

Author: Vasik Rajlich

Date: 06:58:33 01/08/06

Go up one level in this thread


On January 07, 2006 at 09:12:10, Steve Maughan wrote:

>Vasik,
>
>>For pure material-only evaluation, Rybka WinFinder 1.0 in "naive" mode should
>>be an order of magnitude faster than any "real engine".
>
>I've been playing around with WinFinder and have a quick question.  If the naive
>mode is only using material why are the nodes per second so low?  When I set the
>mode to naive I get a nps of, at most, double the normal rate (e.g. 150 knps).
>I would imagine that material only would be much faster (e.g. > 1500 nps) - any
>reason why?  Is the naive mode using piece square tables or pure material only
>(e.g. knight worth 3.3 pawns no matter where it's located)?
>
>Regards,
>
>Steve

Steve,

naive mode uses the following:

pawn = 1.00
knight, bishop = 3.00
rook = 5.00
queen = 9.00

Every score reported by this mode is a multiple of 1.00.

Re. nps, please don't compare this to other engine nps values. Some evals are
still done inside the search, and even without this the Rybka nps would still be
relatively low. The main thing to know is that naive mode runs around 4 or 5
times faster than non-naive.

Best regards,
Vas




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.