Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: 3 trouble-maker positions [discussed previously] Expert opinions needed

Author: blass uri

Date: 18:34:12 05/18/99

Go up one level in this thread



On May 18, 1999 at 13:50:52, Rob Shultz wrote:

>I ran the positions on Junior 5.0 on a PII 450 with 45 MEG hash table to produce
>the following data.
<snip>


How do you get the data to print the analysis of Junior5 here?

I guess that you used  some command like copy fen and Paste and did not copy it
bit by bit by looking at Junior5's analysis copying it to a paper and copying it
from a paper to here(this is the only way that I know to give Junior5's
analysis).

Is it possible to get it from Junior5 program or maybe it is possible to do it
only by using Chessbase7?


Interesting to see that Junior5.0 is probably better in finding the best moves
than Junior5.4 in these positions.

Junior5.4 cannot solve these positions but I think that Junior5.4 is a better
player.
I have the following data about results of Junior5.4 in engine-engine games:
Junior5.4-Fritz5(16 bit) 12:8 in the nunn match (99 minutes per game)
It was 11.5:8.5 in the chessbase match.

Junior5.4 did not do a good result in the nunn match at faster time control(2
minutes per game +15 seconds per move(lost 14.5:5.5) but I am more interested in
longer time control

Junior5.4-crafty16.6 11.5-6.5 in engine-engine games.
I used my pentium200MMX.
Time control  40 moves/2 hours +20 moves/ 1 hour+30 minutes per game.

I used 64 Mb hash tables for both engines for all of these tests.

The openings were the same as the opening of James Walker in his games(1 hour
per game on 2 K6-350)
The results was 9:9 in James walker games(16:14 for Junior5 after 30 games)

I started from the positions when Junior was out of book in the games
In most of the cases Junior was the first side to go different.

In cases that crafty did not use time for going different in the original
games(crafty probably used a book) I stopped the game and started the game from
the position after the first move that crafty played a different move in the
original game.

This test is not the same as James walker because of the following facts:

1)Time control is not the same and the computer is not the same.
I tried to do similiar time control using the facts that my pentium200MMX(win95)
is probably twice slower than the K6-350 and I use one computer and not 2
computers in my engine-engine test.

2)Crafty is not the same as Crafty of James walker and is probably slightly
weaker(I do not think that there is a big difference and I expect Crafty16.6 as
an engine for Junior to get more than 40% against real craty16.6)

3)The engines cannot use the permanent brain when the engines could use the
permanent brain in James walker games because I use only one computer.

I do not tell the engines permanent brain off but it is clear that they do not
use it otherwise I could see Junior showing sometimes big depth when it start to
calculate.

I think it should help Crafty  because the fact that Junior is clearing the hash
tables after every move and crafty does not do it is more significant. but I am
not sure about it.

4)I see that crafty is not clearing the hash tables as Junior's engine because
sometimes I see Crafty goes to depth 10 at 0 seconds but I am not sure if it is
the same as the real crafty.

5) Sometimes Crafty saved time relative to the original games because it started
the game when Junior was out of book and not before Junior was out of book.
Sometimes it was the opposite and Crafty used time to find moves that were book
moves in the original games.

6)The engines are sometimes slowed down for a small part of the games by a
factor of 2-3 during the games.
I do not know the reason for it and maybe I should blame windows95

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.