Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: update after running 7:48 hours with Deep Junior7

Author: Uri Blass

Date: 06:56:20 07/23/02

Go up one level in this thread


On July 23, 2002 at 09:45:35, K. Burcham wrote:

>
>
>Deep Junior7 does not know to count nodes correctly.
>
>The number is 2,957,587 knodes and it means 2,957,587,000 nodes but you can
>trust the number only when the number of nodes is smaller than 2^32.
>
>Uri
>
>26.axb5 axb5 27.Ra5 g6 28.Bb6 Qe8 29.Bd1 h5 30.Nf1 Nf6 31.Nd2
>  ²  (0.67)   Depth: 25   01:30:55  4282117kN
>26.axb5 axb5 27.Ra5 g6 28.Bb6 Qe8 29.Bd1 h5 30.Nf1 Be7 31.R5a2 h4
>  ²  (0.66)   Depth: 26   03:17:03  1734361kN
>26.axb5 axb5 27.Ra5 g6 28.Bb6 Qe8 29.Bd1 h5 30.Qa2 Rxa5 31.Qxa5 Bh6 32.Qa7 Bd2
>33.Ne2 Be1 34.Be3
>  ²  (0.66)   Depth: 27   07:48:23  2957587kN
>
>
>Uri, DJ7 shows the number is 2957587kn. which equals 2,957,587,000.
>but you can see from the eval that DJ dropped a digit between depth 25 and depth
>26. based on the increase between each update it seems DJ dropped a 0.
>if we add a 0, we get 29,575,870,000.
>
>kburcham

No

I am a programmer and I know that movei has also similiar problem.
The number of nodes is calculated as n mode 2^32.

The numbers also suggest that your explanation is not the right explanation.




We have:

Depth: 24   00:38:17  3500682kN
26.axb5 axb5 27.Ra5 g6 28.Bb6 Qe8 29.Bd1 h5 30.Nf1 Nf6 31.Nd2
  ²  (0.67)
Depth: 25   01:30:55  4282117kN


The time is more than twice but the number of nodes is not.

If we assume that deep junior dropped a digit then

03:17:03  1734361kN

means that the number of nodes here is more than 4 times bigger when the time is
only slightly more than twice bigger.

It is not logical.

Uri



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.