New Infinite Loop-6, Ver-qual 2

[Igor Gorelikov]

Geschrieben von:/Posted by: Igor Gorelikov at 02 July 2004 15:56:24:

New Infinite Loop-6, Ver-qual 2
-------------------------------
The 2nd qualifier for new versions is concluded.
To be qualified, a new version needs to take upper place than the best
previous version.
Promoted to New IL-6:
---------------------
Booot 4.0.3
Djinn 0.870
Failed promotion:
-----------------
Ufim 5.00
GreKo 2.80
BlackBishop 0.9.7i
Time control: 30'+3"
Celeron 1GHz 256MB; hash settings is up to 38MB in sum for each engine;
resign is set to 6; few 5-men Nalimov's TBs.
Galis WBTM 0.5.0 is used to arrange the event.

Results
P3 1GHz 256MB, 2004.06.29 - 2004.07.02
1 2 3 4 5 6 7 8 9 0 S R B
-----------------------------------------------------------------------
1 Booot 4.0.3 XX 10 11 1= 11 10 1= 01 == 11 13 1 113.50
2 Ufim 4.02 01 XX 01 =1 1= 01 =0 11 11 11 12= 2-3 101.25
3 Djinn 0.870 00 10 XX =0 11 1= 1= 11 11 11 12= 2-3 91.50
4 Ufim 5.00 0= =0 =1 XX 01 01 11 =1 11 11 12 4 92.00
5 Djinn 0.848 00 0= 00 10 XX 11 11 1= 1= 01 9= 5 68.50
6 GreKo 2.70 01 10 0= 10 00 XX 00 1= 01 == 7 6-7 63.25
7 GreKo 2.80 0= =1 0= 00 00 11 XX 10 1= 00 7 6-7 59.75
8 BlackBishop 0.9.2d 10 00 00 =0 0= 0= 01 XX =1 == 6 8 47.50
9 Booot 3.1 == 00 00 00 0= 10 0= =0 XX 11 5= 9 41.25
0 BlackBishop 0.9.7i 00 00 00 00 10 == 11 == 00 XX 5 10 36.50
-----------------------------------------------------------------------
Games 90/90, +40 -30 =20


Next qualifier:
---------------
New IL-6, Ver-qual 3
95. Butcher 1.38
Butcher 1.47 new version
97. Frenzee 146
Frenzee 159 new version
111. TRACE 1.27
TRACE 1.29 new version
116. Beowulf 2.2
Beowulf 2.3b5 new version
129. KnightX 1.82
KnightX 1.83 new version


All the games of the events reported today can be found at
http://www.digichess.gr/infiniteloop/games/last.zip
====================================================
For games and more info on the Infinite Loop apply to
http://www.digichess.gr/infiniteloop/home.php

[Niyaz Khasanov]

Geschrieben von:/Posted by: Niyaz Khasanov at 02 July 2004 16:29:24:
Als Antwort auf:/In reply to: New Infinite Loop-6, Ver-qual 2 geschrieben von:/posted by: Igor Gorelikov at 02 July 2004 15:56:24:

All the games of the events reported today can be found at
http://www.digichess.gr/infiniteloop/games/last.zip
====================================================

Unfortunately, I have problems with your site for a long time.
Could you send me the last.zip file, please?
My e-mail address: salavata(at)mail.ru
(at) is replaced by @

[Igor Gorelikov]

Geschrieben von:/Posted by: Igor Gorelikov at 02 July 2004 16:48:08:
Als Antwort auf:/In reply to: Re: New Infinite Loop-6, Ver-qual 2 geschrieben von:/posted by: Niyaz Khasanov at 02 July 2004 16:29:24:

[Niyaz Khasanov]

Geschrieben von:/Posted by: Niyaz Khasanov at 02 July 2004 16:49:38:
Als Antwort auf:/In reply to: done (n/t) geschrieben von:/posted by: Igor Gorelikov at 02 July 2004 16:48:08:

[Andreas Herrmann]

Geschrieben von:/Posted by: Andreas Herrmann at 02 July 2004 21:12:13:
Als Antwort auf:/In reply to: New Infinite Loop-6, Ver-qual 2 geschrieben von:/posted by: Igor Gorelikov at 02 July 2004 15:56:24:

New Infinite Loop-6, Ver-qual 2
-------------------------------
The 2nd qualifier for new versions is concluded.
To be qualified, a new version needs to take upper place than the best
previous version.

I'm sure that 0.9.7i is 45 to 50 ELO stronger than the buggy version 0.9.2d. I have playd 1340 games (40/3) with each version against the same opponents and under the same conditions.

BB0.9.7i: 678,5/1340 = 50,6%
BB0.9.2d: 589,5/1340 = 44,0%

BB0.9.7i was just unlucky in your tournament. And i'm sure that you know, that 18 games is not enough to show if one engine is 50 ELO better than another.
I think, it would be better to use always the newer version. Also because no one is interested in results of old and buggy versions. Just my opinion.
best wishes
Andreas




wbholmes.de

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 02 July 2004 22:31:29:
Als Antwort auf:/In reply to: Re: New Infinite Loop-6, Ver-qual 2 geschrieben von:/posted by: Andreas Herrmann at 02 July 2004 21:12:13:

New Infinite Loop-6, Ver-qual 2
-------------------------------
The 2nd qualifier for new versions is concluded.
To be qualified, a new version needs to take upper place than the best
previous version.

I'm sure that 0.9.7i is 45 to 50 ELO stronger than the buggy version 0.9.2d. I have playd 1340 games (40/3) with each version against the same opponents and under the same conditions.

BB0.9.7i: 678,5/1340 = 50,6%
BB0.9.2d: 589,5/1340 = 44,0%

BB0.9.7i was just unlucky in your tournament. And i'm sure that you know, that 18 games is not enough to show if one engine is 50 ELO better than another.
I think, it would be better to use always the newer version. Also because no one is interested in results of old and buggy versions.

The last reason is wrong.
people are interested in results of old and buggy versions in order to compare
with better versions.
There are rules to the tournament and Igor is free to keep his rules.
public movei also did not promote by his rules because oider version that I sent him did better against the baron and the fact that the new version did better against thinker was not enough compensation.
Uri
Uri

[Igor Gorelikov]

Geschrieben von:/Posted by: Igor Gorelikov at 05 July 2004 13:31:14:
Als Antwort auf:/In reply to: Re: New Infinite Loop-6, Ver-qual 2 geschrieben von:/posted by: Andreas Herrmann at 02 July 2004 21:12:13:

New Infinite Loop-6, Ver-qual 2
-------------------------------
The 2nd qualifier for new versions is concluded.
To be qualified, a new version needs to take upper place than the best
previous version.

I'm sure that 0.9.7i is 45 to 50 ELO stronger than the buggy version 0.9.2d. I have playd 1340 games (40/3) with each version against the same opponents and under the same conditions.

BB0.9.7i: 678,5/1340 = 50,6%
BB0.9.2d: 589,5/1340 = 44,0%

BB0.9.7i was just unlucky in your tournament. And i'm sure that you know, that 18 games is not enough to show if one engine is 50 ELO better than another.
I think, it would be better to use always the newer version. Also because no one is interested in results of old and buggy versions. Just my opinion.
best wishes
Andreas

"(40/3)" is a shorter time control vs my 30'+3". At different time controls engines can behavour differently.
I cannot spend more than 18-20 games per engine. One game takes about 50 minutes (on average). You may multiple it by number of games and number of new versions.
In my opinion, 18-20 games is a good compromise.
The main principle behind my Infinite Loop is comparison between different versions. An impressive example of it is Crafty 16.19 which stays as a rock till now against new versions of Crafty.
If "old and buggy versions" show better performance... sorry, I cannot correct it.
Regards,
Igor