FCP: TOP-8 tourney, 40 in 40 in LIVE mode

[Frank Quisinsky]

Hi Jouni,

OK, Komodo game is remis.
Stockfish - Houdini ... remis to 95% or Houdini will win to 5%. I think a clear endgame.

In this case ...

1. Komodo 51,5
2. Stockfish 51,5
3. Houdini 51,0

Best
Frank

[Frank Quisinsky]

Hi there,

Houdini won the game I have to replay vs. Stockfish 2.2.2 in the late endgame (see my other comment, not remis). With the result that Houdini is on two and Stockfish on three. Interesting because Stockfish lost with KNRp vs. KNBR (normaly this endgame is remis).

Final results:

Code: Select all

FCP, 40/40, ponder=on, Q9550   2012

                                1          2          3          4          5          6          7          8          9          0          
1   Komodo 4.0 x64              ********** ½1½½1½½½1½ 10½11½½½1½ 0½½11½½½1½ ½01½0½½½01 ½½½11½0101 ½½½1½1000½ 1½½1½0½0½½ 1½½01½½½½½ 1½½½11½½1½  51.5/90  2273.25
2   Houdini 2.0c x64            ½0½½0½½½0½ ********** 011½½10½1½ 110½1½1½½1 ½01½½10½0½ ½1½10½1½½½ 10½11½½½½½ 0½½1½0½½½0 ½1½½½1½½11 1½011111½½  51.5/90  2238.75
3   Stockfish 2.2.2 JA x64      01½00½½½0½ 100½½01½0½ ********** ½½½½11½01½ 1½½½1½1½½½ ½½000011½½ 111½0½½½½1 ½½010½0½11 11½½½1½1½½ ½1½11111½½  50.5/90
4   Rybka 4 x64 Exp. 42         1½½00½½½0½ 001½0½0½½0 ½½½½00½10½ ********** ½½½½0½½0½0 ½½1111½½0½ ½1½01111½½ ½1½11½0½11 11½11½½½½½ 1½½1½½11½1  49.5/90
5   Stockfish 2.1.1 JA x64 PHQ  ½10½1½½½10 ½10½½01½1½ 0½½½0½0½½½ ½½½½1½½1½1 ********** ½0½000½½11 00½½½10½½0 01½1½0½1½½ ½1½1½1½½½½ 01½1½½1½½½  46.5/90
6   Critter 1.4 x64             ½½½00½1010 ½0½01½0½½½ ½½111100½½ ½½0000½½1½ ½1½111½½00 ********** ½11½½1½0½0 ½1½1½½½1½½ ½1½0½½½½½½ ½½½½½½½101  46.0/90
7   IvanHoe 999946hm x64        ½½½0½0111½ 01½00½½½½½ 000½1½½½½0 ½0½10000½½ 11½½½01½½1 ½00½½0½1½1 ********** 0½1½1½½1½½ ½10½0½0½½0 ½½1½½0½111  43.0/90  1907.00
8   RobboLito 0.10 x64          0½½0½1½1½½ 1½½0½1½½½1 ½½101½1½00 ½0½00½1½00 10½0½1½0½½ ½0½0½½½0½½ 1½0½0½½0½½ ********** ½½½½½½0½½½ ½½½1½11½11  43.0/90  1903.50
9   Rybka 4.1 x64               0½½10½½½½½ ½0½½½0½½00 00½½½0½0½½ 00½00½½½½½ ½0½0½0½½½½ ½0½1½½½½½½ ½01½1½1½½1 ½½½½½½1½½½ ********** ½1½½1½½1½0  40.0/90
10  Chiron 1.1a x64             0½½½00½½0½ 0½100000½½ ½0½00000½½ 0½½0½½00½0 10½0½½0½½½ ½½½½½½½010 ½½0½½1½000 ½½½0½00½00 ½0½½0½½0½1 **********  28.5/90

ELOstat 1.1

Code: Select all

    Program                            Score     %    Av.Op.  Elo    +   -    Draws

  1 Houdini 2.0c x64               :  51.5/ 90  57.2   2945   2996   50  50   52.2 %
  2 Komodo 4.0 x64                 :  51.5/ 90  57.2   2945   2996   48  47   56.7 %
  3 Stockfish 2.2.2 JA x64         :  50.5/ 90  56.1   2946   2989   51  51   50.0 %
  4 Rybka 4 x64 Exp. 42            :  49.5/ 90  55.0   2947   2982   50  50   52.2 %
  5 Stockfish 2.1.1 JA x64 PHQ     :  46.5/ 90  51.7   2949   2961   48  47   56.7 %
  6 Critter 1.4 x64                :  46.0/ 90  51.1   2950   2957   48  48   55.6 %
  7 IvanHoe 999946hm x64           :  43.0/ 90  47.8   2952   2937   50  51   51.1 %
  8 RobboLito 0.10 x64             :  43.0/ 90  47.8   2952   2937   47  47   57.8 %
  9 Rybka 4.1 x64                  :  40.0/ 90  44.4   2954   2916   41  42   66.7 %
 10 Chiron 1.1a x64                :  28.5/ 90  31.7   2964   2830   48  50   52.2 %

It seems that my provider have problems.
So my webpage isn't available so far and I have to add the games a little bit later.

Congratulation to the Komodo team.
But with longer time control it's clear that Houdini, Robbolito, Ivanhoe and Rybka 4.1 lost a bit points and Stockfish and Komodo, also the very strong Rybka 4 Exp. setting is stronger. I think the final results are very clear, not new information for myself.

Best
Frank

[Frank Quisinsky]

Hi Carlos,

yes, only 50% remis quote for Stockfish 2.2.2 JA x64. Good endgame without endgame databases. Also Stockfish 2.1.1. JA PHQ is playing a good tourney.

But we can see that Stockfish can't win so many fast games vs. stronger opponents. Only new for myself is, that Komodo is playing the most interesting short games with nice tactical moves. Rybka 4 Exp. 42 lost the most fast games but perhaps have the best endgame from all what I can see here.

All in all ...
With longer time controls we have only a bit other results as with 40 in 10 (SWCR). This tourney was playing with 40 in 40 = 4x more time, around 170 minutes for one game. Critter 1.4 lost also a bit power in this tourney. Not enough games, but it was only a tourney, not a rating list.

Many remis games, much more if we compare the results with 40 in 10. Also with 101 moves without resign mode the move average is around 12 moves higher as with 40 in 10.

Clear is that with a higher remis quote we need not to many results for a clear rating. Perhaps 300 or 400 games are enough I think. With 40 in 10 are around 800 games enough, with 40 in 3 (my older blitz list) around 1.200 - 1.400 games. Means that the work CEGT do with 40 in 120 is very very good, also with 300-600 games only (really enough).

Have fun with Stockfish, really a very nice engine ... I like it too. Also a very nice tourney from Komodo and Ivanhoe plays a good second half from this tourney. Perhaps the IvanHoe version is a bit stronger as in my tourney. All other results are very clear I think.

Best
Frank

[Frank Quisinsky]

http://www.amateurschach.de/test-tourne ... est-01.htm

Here you can replay all 450 games or download the games (CBH with Shredder GUI comments and PGN without Shredder GUI comments) = 1.2 Mb.

Best
Frank

[Carlos Ylich]

Hello Frank,
I believe that Komodo is the engine stronger pace today (40/40).
Stockfish is also very strong. Your information is very good and
important. Thanks for your great job informing us.
Carlos Ylich

[Sedat Canbaz]

Hi there,

at the moment I don't have many time to redevelop my webpage. Furthermore, I am reading the book from Larry Kaufmann for an review I have interest to create.

But for the moment you can visit a TOP-8 tourney in LIVE mode if you have interest.

4x more time as in SWCR ...

URL:
http://www.amateurschach.de/ftptrigger/test-01.html

Code: Select all

Conditions:
#Time control: 40/40 "repeatedly"
#Game average: 160 minutes
# - 8 games running to the same time
# - 10 games per match, 10 engines = 450 games
# - Tournament time: January 27th, 2012 - February 03th, 2012 
#Resign: OFF
#Ponder: ON
#Learning: OFF
#Endgames: 4-pieces, 32Mb cache
#Opening books: OWN v5.11
#GUI: Shredder Classic 4
#OS: Windows XP Prof. x64 Edition
#Processors: Intel, 4xQ9550 2,83GHz
#Cores: 1 core for each engine
#Hash-Tables: 512Mb

Games are in around one week available for donwload.
Have fun with this little tourney.

Best
Frank

Hello dear Frank,

Many thanks again for organizing this interesting tournament for us

It seems, Komodo (under a such conditions) is a beast !

And i have no patience to see Komodo MP's performance

Greetings,
Sedat

[Frank Quisinsky]

Hi Sedat,

yes, after all what I saw here I am very surprise about Komodo. It seems this engine need really time to find more tactical moves. Absolutley possible that Komodo will win more points with longer time controls but more interesting is ... Komodo played really very nice games in this tourney.

For myself is the SMP version not important. I analyze on Quad systems not with SMP power. In this time I like it to used 4 engines under ChessBase 11 with 1 core (more interesting for myself). But of course for SMP fans could be this version an event.

Best
Frank

[Frank Quisinsky]

Hi Carlos,

no problem.
Thanks for your comments and interest and have fun with all the nice available chess engines.

Best for yourself
Frank

[Sedat Canbaz]

Hi Sedat,

yes, after all what I saw here I am very surprise about Komodo. It seems this engine need really time to find more tactical moves. Absolutley possible that Komodo will win more points with longer time controls but more interesting is ... Komodo played really very nice games in this tourney.

For myself is the SMP version not important. I analyze on Quad systems not with SMP power. In this time I like it to used 4 engines under ChessBase 11 with 1 core (more interesting for myself). But of course for SMP fans could be this version an event.

Best
Frank

Dear Frank,

Honestly i prefer/like mainly testings with maximum performance (with many cores)

Some notes about 1 core ratings,
Yes...i know very well too that we can run multiply testings on Quads,Six core machines...
I mean, we can be produced much more games than Auto232 ratings or creating ratings with maximum cores

But however,the games which are played with maximum performance (max. cores) are more quality than the games with 1 core
In other words,i prefer to drive my auto machine with 150km -200km per hour (on a good highway) than 30km -50km per hour

Normally if one day i will create a rating list on only 1 core engines,then my conditions will be:
-No opening books
-Only 32 bit engines
-No endgames

Note:just only under such conditions we can say:it is accurate rating list (no any advantage ...)

About SWCR,
I like a lot your systematically work,your site is one of my favorites
And it seems SWCR is one of the most accurate Elo ratings
Sad that you stop a such great project,but i have no doubt that you will come with another interesting project



Best,
Sedat