Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Idea for standard NPS test: does this make any sense?

Author: James Swafford

Date: 09:16:10 01/30/01

Go up one level in this thread


On January 30, 2001 at 03:26:58, Dann Corbit wrote:

>On January 30, 2001 at 03:12:19, Jouni Uski wrote:
>
>>Let's take 4 first LCT II position, which are middlegame positions:
>>
>>r3kb1r/3n1pp1/p6p/2pPp2q/Pp2N3/3B2PP/1PQ2P2/R3K2R w KQkq - ; id "POS-01";c0
>>"Chernin - Miles, Tunis 1985, 1.d6!";
>>1k1r3r/pp2qpp1/3b1n1p/3pNQ2/2pP1P2/2N1P3/PP4PP/1K1RR3 b - - ; id "POS-02";c0
>>"Lilienthal - Botvinnik, Moskau 1945, 1...Bb4!";
>>r6k/pp4p1/2p1b3/3pP3/7q/P2B3r/1PP2Q1P/2K1R1R1 w - - ; id "POS-03";c0 "Boissel -
>>Boulard, corr. 1994, 1.Qc5!";
>>1nr5/2rbkppp/p3p3/Np6/2PRPP2/8/PKP1B1PP/3R4 b - - ; id "POS-04";c0 "Kaplan -
>>Kopec, USA 1975, 1...e5!";
>>
>>And let any program to calculate each position 3 minutes and then count average
>>NPS from all four...
>
>NPS of what?
>The move generator?
>Make/unmake?
>Nodes used in search?
>or
>???


:-)
This goes back to the classic:  "How do you count nodes?"

I have use a bench command, used for comparing speed of the
same binary on different platforms, and for analyzing the
effects of source changes.  Comparing my results to the bench
of another program would be almost useless, IMHO.

--
James






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.