Android UCI engine test tournament

As in chess tournaments and matches...
User avatar
AartBik
Posts: 145
Joined: Tue Jun 15, 2010 9:39 pm
Real Name: Aart Bik
Location: Mountain View, CA
Contact:

Re: Android UCI engine test tournament

Post by AartBik » Sun May 22, 2011 8:39 am

Gargamel wrote:Another tournament
Nice. Thanks for including the built-in engine with the "big boys" :-)

User avatar
AartBik
Posts: 145
Joined: Tue Jun 15, 2010 9:39 pm
Real Name: Aart Bik
Location: Mountain View, CA
Contact:

Re: Android UCI engine test tournament

Post by AartBik » Sun May 22, 2011 9:14 pm

A tournament with more games between the top engines, same settings. Congrats to the stockfish team for winning the Android UCI engines tournament!

Code: Select all

                                 1            2            3            4            5            6                          
1   Stockfish 2.0                ***********  50.5 - 49.5  60.0 - 40.0  68.5 - 31.5  69.5 - 30.5  75.0 - 25.0     323.5/500
2   Stockfish 2.1                49.5 - 50.5  ***********  57.0 - 43.0  57.5 - 42.5  70.5 - 29.5  82.0 - 18.0     316.5/500
3   RobboLito 0.085e4l           40.0 - 60.0  43.0 - 57.0  ***********  49.5 - 50.5  58.5 - 41.5  59.5 - 40.5     250.5/500
4   RobboLito 0.085g3l           31.5 - 68.5  42.5 - 57.5  50.5 - 49.5  ***********  56.0 - 44.0  59.5 - 40.5     240.0/500
5   IvanHoe-Beta version 999947c 30.5 - 69.5  29.5 - 70.5  41.5 - 58.5  44.0 - 56.0  ***********  51.0 - 49.0     196.5/500
6   Komodo32 1.3 JA              25.0 - 75.0  18.0 - 82.0  40.5 - 59.5  40.5 - 59.5  49.0 - 51.0  ***********     173.0/500

chesskiller
Posts: 5
Joined: Tue May 03, 2011 10:51 pm

Re: Android UCI engine test tournament

Post by chesskiller » Wed May 25, 2011 6:09 pm

Thanks aart. Nice job. Can anybody tell the elo difference between say droidfish (stochfish 2.1) on a 1 GHz Qualcomm QSD8250 Snapdragon, and stockfish on say i7-2600 (no overclocking)?
Thanks

User923005
Posts: 616
Joined: Thu May 19, 2011 1:35 am

Re: Android UCI engine test tournament

Post by User923005 » Thu May 26, 2011 11:55 pm

AartBik wrote:Test tournament with different Android versions of Miguel Ballicora's Gaviota and Michel Van den Bergh's GNU chess.

Code: Select all

                           1          2          3          4          5          
1   GNU Chess 5.07.170.7b  ********** 000½10011½ 11½1110110 1100½01001 0111101111  24.0/40
2   gaviota v0.80.0.107    111½01100½ ********** ½010110100 0100½11110 ½101000011  20.5/40
3   GNU Chess 5.07.153.3b  00½0001001 ½101001011 ********** 0101010½01 0111111½10  20.0/40
4   gaviota v0.82-beta2    0011½10110 1011½00001 1010101½10 ********** 00110001½0  19.0/40
5   gaviota v0.83          1000010000 ½010111100 1000000½01 11001110½1 **********  16.5/40
This one puzzles me because Gaviota v0.83 is much stronger than any Gaviota predecessors

User avatar
AartBik
Posts: 145
Joined: Tue Jun 15, 2010 9:39 pm
Real Name: Aart Bik
Location: Mountain View, CA
Contact:

Re: Android UCI engine test tournament

Post by AartBik » Fri May 27, 2011 12:02 am

User923005 wrote:This one puzzles me because Gaviota v0.83 is much stronger than any Gaviota predecessors
In none of my tests (including a few private ones) did the latest Gaviota outperform older versions. Perhaps this is because I do most testing with fixed time (1 second per move), rather than the more commonly tested time controls?

User923005
Posts: 616
Joined: Thu May 19, 2011 1:35 am

Re: Android UCI engine test tournament

Post by User923005 » Fri May 27, 2011 1:54 am

AartBik wrote:
User923005 wrote:This one puzzles me because Gaviota v0.83 is much stronger than any Gaviota predecessors
In none of my tests (including a few private ones) did the latest Gaviota outperform older versions. Perhaps this is because I do most testing with fixed time (1 second per move), rather than the more commonly tested time controls?
Not sure why.

Code: Select all

http://www.husvankempen.de/nunn/40_4_Ratinglist/40_4_AllVersion/rangliste.html
no Program Elo + - Games Score Av.Op. Draws 
621 Gaviota 0.83 x64 1CPU 2636 19 19 1000 61.4% 2555 22.2% 
748 Gaviota 0.80 x64 1CPU 2553 18 18 1200 54.9% 2518 20.2% 
856 Gaviota 0.74.41 x64 2469 25 25 600 51.0% 2462 22.3% 
Top two versions here are probably just due to spelling error and should be combinedL:

Code: Select all

http://www.husvankempen.de/nunn/40_40%20Rating%20List/40_40%20All%20Versions/rangliste.html
no Program Elo + - Games Score Av.Op. Draws 
625 Gaviota 0.83 x64 1CPU 2606 22 22 649 47.0% 2627 29.9% 
647 Gaviota v0.83 x64 1CPU 2593 81 81 52 41.3% 2653 28.8% 
705 Gaviota 0.80 x64 1CPU 2546 20 20 806 49.4% 2550 28.5% 
796 Gaviota 0.74 x64 1CPU 2469 29 29 467 55.6% 2430 18.6% 
800 Gaviota 0.75 x64 1CPU 2464 25 25 543 48.6% 2473 25.0% 
844 Gaviota 0.74 w32 2407 30 30 400 50.4% 2404 22.2% 

User avatar
AartBik
Posts: 145
Joined: Tue Jun 15, 2010 9:39 pm
Real Name: Aart Bik
Location: Mountain View, CA
Contact:

Re: Android UCI engine test tournament

Post by AartBik » Sun May 29, 2011 7:04 am

Tournament with 100 games from both sides of a fixed opening book. Still 1 second per move, but now the latest version performs best.

Code: Select all

Chess for Android Tournament  May 2011

                         1            2            3                                                               
1   gaviota v0.83        **           55.5 - 44.5  55.0 - 45.   110.5/200
2   gaviota v0.80.0.107  44.5 - 55.5  **           61.5 - 38.5  106.0/200
3   gaviota v0.82-beta2  45.0 - 55.0  38.5 - 61.5  **            83.5/200

User avatar
thorstenczub
Posts: 593
Joined: Wed Jun 09, 2010 12:51 pm
Real Name: Thorsten Czub
Location: United States of Europe, germany, NRW, Lünen
Contact:

Re: Android UCI engine test tournament

Post by thorstenczub » Sun May 29, 2011 1:51 pm

thanks aart.

User avatar
AartBik
Posts: 145
Joined: Tue Jun 15, 2010 9:39 pm
Real Name: Aart Bik
Location: Mountain View, CA
Contact:

Re: Android UCI engine test tournament

Post by AartBik » Sat Jun 11, 2011 2:47 am

Congrats to Don Dailey and Larry Kaufman for releasing an impressive new version of Komodo! Here are the results of a quick one-second per move tournament, 32MB hash, between two available Android versions on a Nexus S.

Code: Select all

1   Komodo32 2 AB     +58/-12/=30 73.00%   73.0/100
2   Komodo32 1.3 JA   +12/-58/=30 27.00%   27.0/100
And the results of a tournament on a Nexus One between top engines.

Code: Select all

                        1            2            3                                   
1   Stockfish 2.0       ***          48.5 - 51.5  59.0 - 41.0    107.5/200
2   RobboLito 0.085e4l  51.5 - 48.5  **           50.0 - 50.0    101.5/200
3   Komodo32 2.01 AB    41.0 - 59.0  50.0 - 50.0  **              91.0/200

Post Reply