Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: "Gambit Rebel" settings, Re: I am sure..

Author: Christophe Theron

Date: 11:50:33 12/07/00

Go up one level in this thread


On December 07, 2000 at 11:19:10, Eelco de Groot wrote:

>
>Thanks, Christophe! I didn't really think that you would do that.
>Congratulations with your new program! Everything seems to be working fine on my
>Celeron.
>
>I changed the lay-out a bit, shrunk the moves-list a little so that the
>histogram could be put under it. Couldn't find that at first, I believe I used
>'Arrange Icons' to locate it and other windows that are sometimes minimized. The
>colour of the board is now dark green for the dark squares and a specially
>composed honey yellow for the light squares which makes the board stand out a
>little more. The statistics window I usually make a lot larger after some time
>so that I can copy the variations in a Wordpad document. What I liked a lot was
>the possibility of copying the game.pgn to clipboard which is directy accessible
>from ECTool.



Yes, you can exchange both PGN and EPD with the clipboard and that's a GREAT
feature added by our friend Lex (author of the GUI).

Especially useful to copy games and positions from CCC for example, and I use it
every day!




> In ECTool I can then switch on a Rebel engine or two as well with
>the full game details directly copied. Easy! Then I copy the ECTool board as a
>bitmap to the Wordpad document too, full color, no chess-book can match that!
>Or from Rebel Tiger there is a little graphical board font too which can be used
>as a little diagram. Now all I need is grandmaster insight to go with the
>analysis!
>
>I believe I did run into Uri's 'infinite time' bug, at least after about 7 hours
>the NPS counter and the total NPS right at the bottom of Tiger's window were
>showing negative numbers then. Maybe the engine was still working because the
>NPS was still changing. I don't really know, Tiger was already eighteen plies
>deep at the time!



The negative NPS display does not hurt. I agree it is not nice, but it is not
going to cause problem.

There is indeed a bug in the infinite analysis: it will stop after several
hours. We are going to provide a fix for this.




> Try that with Q5T, there there seems to be a natural limit of
>seventeen plies, maybe the extensions of Deeper Search grow too big then. But I
>didn't have enough patience to see if the move that Tiger was contemplating at
>the time would change anymore as an indication that the engine was still
>running. What I did notice is that Tiger is sometimes very modest in asking CPU
>time, with normal engine priority. I'm using Windows 98 Second Edition. There
>was already a DOS box with Century 2.0 running when I started up Tiger 13 which
>went very well. The NPS is normally around 110000 but did not go much higher
>than 10000 now as Rebel was first. After closing down the Rebel engine in ECTool
>the NPS is climbing with about 20 nodes per second, per second (NPSPS or
>something like that). As I am writing this it is up to about 50000 and climbing
>about 10 positions per second each second.


That's normal because the NPS number is computed from total number of nodes seen
divided by total time. Your NPS has gone back to normal, but the NPS displayed
is not the number of nodes computed at the last second. It is the average NPS
since the beginning of the search. Since your search has been slowed down
before, the NPS says it.

The NPS would be instantly back to normal if you stop the search and start it
again.



> Maybe it is more aggressive in asking
>CPU time when playing a game, at the moment it is in analysis mode and infinite
>time. But I thought that for people playing Blitz on the Internet maybe it might
>be an idea to keep an eye on the NPS counter, to see if Windows is giving Tiger
>enough time and if the 'Engine Priority' should not be changed to "high" then if
>necessary in View -> Properties -> Chess Engine -> Advanced.



Engine priority is only useful for very slow computers (486). It will not help
on faster computers.



>Just some observations, thanks Christophe for your programming!


Thanks for your comments.



    Christophe



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.