Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Deep Junior - Huebner, analysis of Deep Junior move later haha

Author: Mogens Larsen

Date: 15:17:58 07/08/00

Go up one level in this thread


On July 08, 2000 at 15:44:33, Chessfun wrote:

>In terms of NPS definately yes. On my Cel about 50% slower.

That's a very big difference. Maybe positions were Deep Junior relies on
multithreading. I think that is the correct term for dual cpu optimized search,
even though I'm sure about the exact definition.

>IMHO Yes. Seems to me that Deep Junior and Junior 6 are not
>the same program with one just set-up to optimize a dual cpu
>system. I have looked at other games in the past and the evals
>never match at a given depth.

There's probably some multithreading capabilities of Deep Junior that is hurt
significantly by using a single cpu. Are you going to buy a dual machine to
unleash that monster?

>Example:
>r4rk1/p1nqbppp/1pn1p3/P1PpP3/8/BPPQ1N2/3N1PPP/R3R2K b - - 0 1
>
>First Deep Junior.
>DEEP JUNIOR 6 - Huebner,R
>
>Analysis by Deep Junior 6.0:
>20...bxc5 21.Ne4 Rfb8 22.Nfg5 g6 23.Bxc5 Bxg5 24.Nxg5 Rxb3
>  ²  (0.39)   depth: 16   00:07:22  107973kN
>
>Then Junior 6a.
>Analysis by Junior 6.1:
>
>20...bxc5 21.Ne4 Rfb8 22.Nxc5 Bxc5 23.Bxc5 Rxb3 24.Qc2 Rb5 25.Bd6 Rxa5 26.Ng5 g6
>27.Rxa5 Nxa5
>  ²  (0.27)   Depth: 16   00:06:42  106120kN
>
>(Bird, Scarb 08.07.2000)

Deep Junior averages about 244Knps vs. 264Knps for Junior 6.0 judging from your
data, but the result is clearly different, so the difference must be larger that
the actual nps gap.

Thank you for the interesting data.

Best wishes...
Mogens



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.