OpenChess

here:

http://www.inwoba.de

you can follow the tourney ratings.

Bye
Ingo

I look forward to seeing the final numbers in these two.

Deep Rybka 4.1 SSE42 - Houdini 1.5a (3012)
Deep Rybka 4.1 SSE42 - Stockfish 2.01 JA (2920)

Hi Ted,

Ted Summers wrote:I look forward to seeing the final numbers in these two.

Deep Rybka 4.1 SSE42 - Houdini 1.5a (3012)
Deep Rybka 4.1 SSE42 - Stockfish 2.01 JA (2920)

I know that there is always some interest in these individual matches, but actually a result out of 100 games is meaningless.

I personaly do not look to any 'A vs B' matches anymore. The overall result is the only thing that matters.

Bye
Ingo

Didn't want to create a new thread over an unimportant test. Also, maybe I screwed up something -- weird result even if it's only 50 games:

Not much change. Still plays at the same level.

So bug fixes don't necessarily contribute to some sort of efficiency which might improve speed/time management.

kingliveson wrote:weird result even if it's only 50 games:

Once again, nothing weird... yet!
Do the statistics/maths: S.D. is approx. 2.5 so the 95% error bar includes DR-IH 25.5-24.5

Better still, if you really find your result "weird", why not seek confirmation by adding 50 or 150 more games?

Rybka 4.1

Deep Rybka 4.1 SSE42 - Houdini 1.5a (3012) 42.5 - 57.5 42.50% Perf=2960
Deep Rybka 4.1 SSE42 - Stockfish 2.01 JA (2920) 56.0 - 44.0 56.00% Perf=2961
Deep Rybka 4.1 SSE42 - Critter 0.90 SSE42 (2895) 63.5 - 36.5 63.50% Perf=2991
Deep Rybka 4.1 SSE42 - Komodo64 1.3 JA (2830) 72.0 - 28.0 72.00% Perf=2994
Deep Rybka 4.1 SSE42 - Naum 4.2 (2824) 72.5 - 27.5 72.50% Perf=2992
Deep Rybka 4.1 SSE42 - Deep Shredder 12 (2800) 79.0 - 21.0 79.00% Perf=3030
Deep Rybka 4.1 SSE42 - Gull 1.2 (2793) 77.0 - 23.0 77.00% Perf=3002
Deep Rybka 4.1 SSE42 - Deep Sjeng c't 2010 32b (2781) 69.0 - 31.0 69.00% Perf=2919
Deep Rybka 4.1 SSE42 - Spike 1.4 (2779) 72.0 - 28.0 72.00% Perf=2943
Deep Rybka 4.1 SSE42 - Protector 1.4.0 x64 (2759) 79.0 - 21.0 79.00% Perf=2989
Deep Rybka 4.1 SSE42 - spark-1.0 SSE42 (2756) 73.5 - 26.5 73.50% Perf=2933
Deep Rybka 4.1 SSE42 - HIARCS 13.2 MP 32b (2752) 81.0 - 19.0 81.00% Perf=3003
Deep Rybka 4.1 SSE42 - Zappa Mexico II (2714) 80.5 - 19.5 80.50% Perf=2960
Deep Rybka 4.1 SSE42 - Deep Onno 1-2-70 (2683) 81.5 - 18.5 81.50% Perf=2940
Deep Rybka 4.1 SSE42 - Hannibal 1.0a (2683) 84.5 - 15.5 84.50% Perf=2977
Deep Rybka 4.1 SSE42 - Deep Junior 12 (2682) 84.0 - 16.0 84.00% Perf=2970
Deep Rybka 4.1 SSE42 - Toga II 1.4 beta5c BB (2670) 87.5 - 12.5 87.50% Perf=3008
Deep Rybka 4.1 SSE42 - Umko 1.1 SSE42 (2632) 86.0 - 14.0 86.00% Perf=2947
Deep Rybka 4.1 SSE42 - Loop 13.6/2007 (2616) 86.0 - 14.0 86.00% Perf=2931
Deep Rybka 4.1 SSE42 - Jonny 4.00 32b (2616) 94.5 - 5.5 94.50% Perf=3110
1521.5 - 478.5 76.07% Perf=2959



2000 out of 2000 games played

Hello

The final result is online: http://www.inwoba.de

Bye
Ingo

IWB wrote:Hello

The final result is online: http://www.inwoba.de

Bye
Ingo

You should consider testing IvanHoe being an "original" of the Ippolit series. You can find all the versions here: http://chess.cygnitec.com/engine/ivanhoe/

And also, make available PGNs of your tests because it's the right thing to do.

Didn't want to create a new thread over an unimportant test. Also, maybe I screwed up something -- weird result even if it's only 50 games:

Code: Select all

Rybka vs IvanHoe 1+0  
                                    
1   IvanHoe 9.49b x64          +63  +17/=25/-8 59.00%   29.5/50
2   Deep Rybka 4.1 SSE42 x64   -63  +8/=25/-17 41.00%   20.5/50

After 200 rounds, this is more like it -- there might have been issue somewhere, I dont believe it was the limited number of games.