Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The new version of Junior is significantly better

Author: blass uri

Date: 06:38:09 07/27/98

Go up one level in this thread



On July 27, 1998 at 06:15:49, Don Dailey wrote:

>On July 26, 1998 at 23:14:45, blass uri wrote:
>
>>I tested the new version of Junior Amir Ban sent me
>>and I did a nunn match between Junior and fritz5
>>
>>I used pentium200MMX
>>
>>Junior5 won the nunn match 12:8
>>(I gave the programs 10 minutes per game+10 seconds per move).
>>I gave every program 12,000Kbytes hash tables(I cannot give exactly 12000 so the
>>computer made it 11,9..Kbytes
>>
>>I think the main reason Junior5 won fitz5 is that Junior was better in tactics.
>>
>>before this match I did a nunn match with 10 minutes per game without 10 seconds
>>per move(both sides used near 10,000Kbytes hash tables).
>>In this match fritz5 won 12:8 but when I looked at the games I saw that
>>in 1 of the games Junior lost in time at move 156 in a draw position.
>>fritz has only king and bishop when Junior had king bishop and a pawn.
>>This is the reason I decided to give the programs more 10 seconds per move.
>>
>>This is a big improvement because Junior4.6 lost to fritz5 14.5:5.5 in 10
>>minutes per move and if I added 10 seconds to every move the result was not
>>better.
>>
>>Uri
>
>Hi Uri,
>
>You should run the match with many more games, 20 games is not nearly
>enough to claim Junior is better at this time control, it's only an
>indication of this.  Especially with a score of 12 to 8.  I have seen
>results much better than this with even more games turn completely
>around.  If 2 games had been won by Fritz the match would have been
>tied.  I'll bet a lot of games could have gone either way.  To see
>what I mean, flip a coin 20 times and note the scores.  Do this a
>few times and note how often the results are exactly 10 tails and
>10 heads, not very often.  Most of the time you will see a few
>extra heads or a few extra tails.  The single most likely result
>will be 10-10 but this will still be a fraction of the time.

The nunn  match has only 20 games.
It is a symmetric match without opening book from 10 known positions
every position is played twice(one time fritz is white ans 1 time Junior is
white)

I did not say the new Junior is significantly better than fritz5 but
significantly better than the old one(the old one lost 14.5:5.5 in the nunn
match)

Uri
>
>- Don



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.