Android tournament with longer time controls

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:

Android tournament with longer time controls

Post by AartBik » Sun Oct 23, 2011 12:53 am

Below the results of a qualification tournament on Chess for Android from both sides of the first Nunn position at one-second per move on a Nexus One (1GHz Snapdragon). Where applicable, UCI engines were given an 8MB hash table and 1MB endgame tablebases cache (the complete 3- and 4-piece Nalimov, Gaviota, Scorpio, and Robbobases endgame tablebases reside on SD card). XBoard engines play at their default settings. Since the Nexus One is a single-core device, "pondering" was disabled

The qualification results are used to split the engines into separate groups. Engines in each group will play each other at longer time controls, starting with the lowest group. The number one in each group will promote to the next group.

All games can be downloaded from the tournament page.

Code: Select all

                                    1  2  3  4  5  6  7  8  9  0  1  2  3  4  5  6  7  8  9  0  1  2  3  4  5  6  7  8  
1   Critter 1.2 32-bit              ** ½0 ½1 10 11 11 11 11 11 1½ 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11  50.5/54
2   Stockfish 2.1.1 JA              ½1 ** 01 0½ 11 11 11 01 11 11 ½1 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11  49.5/54
3   RobboLito 0.085g3l x86          ½0 10 ** 01 10 11 11 11 11 11 01 11 11 11 ½1 11 11 11 11 11 11 11 11 11 11 11 11 11  48.0/54
4   Komodo32 2AB                    01 1½ 10 ** 01 ½½ 11 10 11 11 11 11 01 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11  47.5/54
5   IvanHoe-Beta version 999947c-r  00 00 01 10 ** 11 11 1½ 10 11 11 ½1 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11  46.0/54
6   crafty                          00 00 00 ½½ 00 ** 01 11 01 ½1 ½½ ½1 11 ½1 ½1 ½1 11 1½ 11 11 11 11 11 11 11 11 11 11  39.0/54
7   Toga II 1.4.1SE                 00 00 00 00 00 10 ** 01 11 11 1½ ½1 1½ 11 01 ½½ ½1 11 11 11 11 11 11 11 11 11 11 11  38.0/54
8   scorpio                         00 10 00 01 0½ 00 10 ** 0½ ½½ 11 01 ½1 1½ 11 10 11 10 01 11 11 11 11 ½1 11 11 11 11  35.5/54
9   GarboChess3                     00 00 00 00 01 10 00 1½ ** 00 01 10 11 01 ½½ 11 11 1½ 11 ½1 11 11 11 11 11 11 11 11  34.5/54
10  Rotor 0.6a                      0½ 00 00 00 00 ½0 00 ½½ 11 ** ½½ 01 11 10 1½ ½½ 01 10 11 11 11 11 11 11 11 11 11 11  33.5/54
11  GNU Chess 5.07.153.3b-32        00 ½0 10 00 00 ½½ 0½ 00 10 ½½ ** 10 11 01 ½½ 11 ½½ 1½ ½1 11 11 1½ 1½ 11 11 11 11 11  33.0/54
12  Pepito v1.59                    00 00 00 00 ½0 ½0 ½0 10 01 10 01 ** ½0 ½½ 11 ½0 ½0 1½ 11 11 01 11 11 1½ 11 11 11 11  30.0/54
13  Daydreamer 1.75 JA              00 00 00 10 00 00 0½ ½0 00 00 00 ½1 ** 11 10 11 11 ½0 10 11 ½1 11 11 11 11 11 11 11  29.5/54
14  gaviota v0.84                   00 00 00 00 00 ½0 00 0½ 10 01 10 ½½ 00 ** ½½ 10 1½ 11 1½ 01 11 11 11 11 11 11 11 11  29.0/54
15  Diablo 0.5.1b JA                00 00 ½0 00 00 ½0 10 00 ½½ 0½ ½½ 00 01 ½½ ** 10 ½1 ½½ 0½ 11 11 11 11 11 11 11 11 11  28.5/54
16  sloppy                          00 00 00 00 00 ½0 ½½ 01 00 ½½ 00 ½1 00 01 01 ** ½0 ½1 11 11 11 1½ 11 11 11 11 11 11  28.5/54
17  RedQueen 0.9.8 JA               00 00 00 00 00 00 ½0 00 00 10 ½½ ½1 00 0½ ½0 ½1 ** 11 01 11 11 1½ 11 11 11 11 11 11  27.0/54
18  Sungorus 1.4 JA                 00 00 00 00 00 0½ 00 01 0½ 01 0½ 0½ ½1 00 ½½ ½0 00 ** 11 11 1½ 1½ 11 11 11 11 11 11  26.0/54
19  GreKo 8.0                       00 00 00 00 00 00 00 10 00 00 ½0 00 01 0½ 1½ 00 10 00 ** ½1 11 11 10 11 11 11 11 11  22.0/54
20  DoubleCheck 1.3                 00 00 00 00 00 00 00 00 ½0 00 00 00 00 10 00 00 00 00 ½0 ** 01 01 1½ 11 11 11 11 11  15.5/54
21  Jazz v5.01 JA                   00 00 00 00 00 00 00 00 00 00 00 10 ½0 00 00 00 00 0½ 00 10 ** 11 00 1½ 11 11 11 11  14.5/54
22  BikJump v2.1P                   00 00 00 00 00 00 00 00 00 00 0½ 00 00 00 00 0½ 0½ 0½ 00 10 00 ** 11 1½ 11 ½1 11 11  14.0/54
23  ZCT-0.3.2500                    00 00 00 00 00 00 00 00 00 00 0½ 00 00 00 00 00 00 00 01 0½ 11 00 ** 11 11 11 11 11  14.0/54
24  tscp181                         00 00 00 00 00 00 00 ½0 00 00 00 0½ 00 00 00 00 00 00 00 00 0½ 0½ 00 ** 1½ 11 11 11   9.5/54
25  umax48w                         00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0½ ** ½1 10 11   5.0/54
26  fairymax48q                     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ½0 00 00 ½0 ** 1½ 11   4.5/54
27  Chess for Android               00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 01 0½ ** 11   3.5/54
28  Simplex 0.9.7 rev 180           00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 **   0.0/54
[/size]

User avatar
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Android tournament with longer time controls

Post by kingliveson » Sun Oct 23, 2011 2:42 am

Am a bit surprise to see RobboLito ahead of the latest IvanHoe, though number of games are small. Thanks for your work.
PAWN : Knight >> Bishop >> Rook >>Queen

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

Re: Android tournament with longer time controls

Post by AartBik » Sun Oct 23, 2011 2:48 am

kingliveson wrote:Am a bit surprise to see RobboLito ahead of the latest IvanHoe, though number of games are small. Thanks for your work.
Yes, this was just the qualification tournament. In the "real" tournament, engines will play all other engines in their group from both sides of all positions of the Nunn opening suite at longer time control. That will give a more serious indication of strength. The tournament in group F has started.

pj+
Posts: 20
Joined: Mon Feb 14, 2011 12:22 am

Re: Android tournament with longer time controls

Post by pj+ » Sun Oct 23, 2011 6:58 am

kingliveson wrote:Am a bit surprise to see RobboLito ahead of the latest IvanHoe, though number of games are small. Thanks for your work.
Is 1 core Ivanhoe coming ahead of Robbolito on linux? I thought that for distributed windows binaries of ivanhoe a little bit more tuning was done in comparison to ippolit wiki.

User avatar
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Android tournament with longer time controls

Post by kingliveson » Sun Oct 23, 2011 5:16 pm

pj+ wrote:
kingliveson wrote:Am a bit surprise to see RobboLito ahead of the latest IvanHoe, though number of games are small. Thanks for your work.
Is 1 core Ivanhoe coming ahead of Robbolito on linux? I thought that for distributed windows binaries of ivanhoe a little bit more tuning was done in comparison to ippolit wiki.

You are probably talking about some modification of IvanHoe made by different compilers. The code for both OSes are the same.


@ AartBik, looking forward to the final standing...
PAWN : Knight >> Bishop >> Rook >>Queen

pj+
Posts: 20
Joined: Mon Feb 14, 2011 12:22 am

Re: Android tournament with longer time controls

Post by pj+ » Mon Oct 24, 2011 6:00 pm

kingliveson wrote:
pj+ wrote:
kingliveson wrote:Am a bit surprise to see RobboLito ahead of the latest IvanHoe, though number of games are small. Thanks for your work.
Is 1 core Ivanhoe coming ahead of Robbolito on linux? I thought that for distributed windows binaries of ivanhoe a little bit more tuning was done in comparison to ippolit wiki.
I am saying it seems that assumption that Ivanhoe is ahead of Robbolito on 1 core came from tests made with windows executables. I think that these windows executables were compiled from modified code. Or there are unmodified windows executables somewhere in rating lists? If there are no tests with unmodified windows executables, I am not sure that Ivanhoe will be stronger than robbolito on linux. As for arm (still linux) Ivanhoe seems to be much slower than robbolito

User avatar
kingliveson
Posts: 1388
Joined: Thu Jun 10, 2010 1:22 am
Real Name: Franklin Titus
Location: 28°32'1"N 81°22'33"W

Re: Android tournament with longer time controls

Post by kingliveson » Wed Oct 26, 2011 8:55 pm

@ AartBik, It was not my intention to hijack this thread, so my apologies.



@ pj+,

Windows

Code: Select all

    Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 IvanHoe 9.47b x64              : 3009   42  41   100    52.5 %   2991   63.0 %
  2 RobboLito 0.085g3 x64          : 2991   41  42   100    47.5 %   3009   63.0 %
Linux

Code: Select all

   Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 IvanHoe 9.47b x64              : 3012   36  36   131    53.4 %   2988   64.1 %
  2 RobboLito 0.085g3l x86_64      : 2988   36  36   131    46.6 %   3012   64.1 %
The first 31 games used neither opening book nor positions while the next 100 used the same 50 positions as Windows...and with enough games, the Elo difference should be about 30 +/- 5.

pj+ wrote: I am saying it seems that assumption that Ivanhoe is ahead of Robbolito on 1 core came from tests made with windows executables. I think that these windows executables were compiled from modified code. Or there are unmodified windows executables somewhere in rating lists? If there are no tests with unmodified windows executables, I am not sure that Ivanhoe will be stronger than robbolito on linux. As for arm (still linux) Ivanhoe seems to be much slower than robbolito
Attachments
20111026_Lin_ro_iv_1m_131r.pgn
(436.79 KiB) Downloaded 232 times
20111026_Win_ro_iv_1m_100r.pgn
(535.9 KiB) Downloaded 219 times
PAWN : Knight >> Bishop >> Rook >>Queen

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

Re: Android tournament with longer time controls

Post by AartBik » Sat Oct 29, 2011 3:27 am

Group F has just finished a complete thirty seconds-per-move tournament, played from both sides of all positions in the Nunn opening suite under the conditions described earlier.

Congratulations to Evert Glebbeek for the Jazz's promotion to Group E.

All games can be downloaded from the tournament page.

Code: Select all

                 1         2          3         4         5                    
1 Jazz v5.01 JA     *      15.5-4.5   18.0-2.0  18.5-1.5  19.5-0.5  71.5/80  <= PROMOTED TO E
2 ZCT-0.3.2500   4.5-15.5      *      13.5-6.5  17.0-3.0  16.0-4.0  51.0/80
3 BikJump v2.1   2.0-18.0   6.5-13.5      *     12.5-7.5  12.0-8.0  33.0/80
4 tscp181        1.5-18.5   3.0-17.0   7.5-12.5     *     12.5-7.5  24.5/80
5 umax48w        0.5-19.5   4.0-16.0   8.0-12.0  7.5-12.5     *     20.0/80

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

Re: Android tournament with longer time controls

Post by AartBik » Sat Oct 29, 2011 3:29 am

kingliveson wrote:@ AartBik, It was not my intention to hijack this thread, so my apologies.
No worries. We are discussing chess, that is what counts :-)

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

Re: Android tournament with longer time controls

Post by AartBik » Sun Nov 06, 2011 9:35 pm

After burning cycles on my Nexus One for many days, Group E has just finished a complete thirty seconds-per-move tournament in Chess for Android, played from both sides of all positions in the Nunn opening suite under the conditions described earlier.

Congratulations to Ben-Hur Carlos Vieira Langoni Jr for RedQueen's promotion to Group D.

All games can be downloaded from the tournament page.

Code: Select all

                    1         2         3         4         5                    
1 RedQueen 0.9.8 JA     *     10.0-10.0 14.5-5.5  17.0-3.0  19.5-0.5  61.0/80  => PROMOTED TO D
2 greko8.2          10.0-10.0     *     14.5-5.5  16.0-4.0  18.5-1.5  59.0/80
3 Sungorus 1.4 JA    5.5-14.5  5.5-14.5     *     13.0-7.0  16.0-4.0  40.0/80
4 Jazz v5.01 JA      3.0-17.0  4.0-16.0  7.0-13.0      *    14.5-5.5  28.5/80
5 DoubleCheck 1.3    0.5-19.5  1.5-18.5  4.0-16.0  5.5-14.5     *     11.5/80

Post Reply