Jim Ablett's Android Chess Engine Bundle stress test

[abik]

Jim Ablett very kindly posted an Android Chess Engine Bundle with various UCI and XBoard engines compiled for ARM, which can be imported in Chess for Android.

Code: Select all

UCI:
daydreamer-175-ja  greko-82-ja      redqueen-098-ja  stockfish-211-ja   diablo-051b-ja
jazz-501-ja        rotor-04-ja      sungorus-14-ja   doublecheck-21-ja  komodo-13-ja
simplex-098-ja     garbochess-3-ja  pepito-159-ja    sissa-20-ja

XBOARD:
uzz-os-11-27-08    kmtchess-121-ja       scidlet-361-b2-ja   chess-one-201-ja         knightcap-36-ja
sjeng-112-ja       crafty-234-ja         mizar-030-ja        sloppy-023-ja            exchess-054-beta-ja
mskcp-145-ja       surprise-43-beta13-ja gfc-08-rev5-ja      natwarlal-014-ja         toledo-nanochess-2010-ja
gk-090-ja          olithink-530-ja       zzzzzz-351-ja       gullydeckel-215-pl10-ja  phalanx-xxii-ja

Because the Win/XBoard protocol feature is still relatively new in Chess for Android, I conducted a test tournament between all the XBoard engines (except gfc-08-rev5-ja which often claims a mate without actually making the move; I am not sure if the standard allows this, but it seems weird that the GUI should verify the claim). During this tournament I found a few opportunities for improving the GUI's handling of the protocol, which will be released soon.

Games available on request.

Code: Select all

                              1  2  3  4  5  6  7  8  9  0  1  2  3  4  5  6  7  8  9  0  
1   olithink-530-ja           ** ½1 11 11 1½ 11 ½1 11 10 11 11 01 11 11 11 11 11 11 11 11  34.5/38
2   crafty-234-ja             ½0 ** 1½ 10 11 11 1½ 11 11 1½ 0½ ½1 11 11 11 11 11 11 11 11  32.0/38
3   phalanx-xxii-ja           00 0½ ** ½1 ½1 11 11 10 11 11 ½1 01 11 01 11 ½1 11 11 11 11  29.5/38
4   chess-one-201-ja          00 01 ½0 ** 1½ 10 ½1 ½1 ½1 11 11 ½1 11 10 11 11 11 11 11 11  29.0/38
5   gullydeckel-215-pl10-ja   0½ 00 ½0 0½ ** 10 1½ 01 11 0½ 11 11 01 11 10 11 11 11 11 11  25.5/38  383.75
6   sloppy-023-ja             00 00 00 01 01 ** 10 0½ 11 01 11 01 11 11 11 11 11 11 11 11  25.5/38  368.50
7   kmtchess-121-ja           ½0 0½ 00 ½0 0½ 01 ** ½1 10 11 11 01 10 10 10 ½1 11 11 11 11  23.0/38  343.75
8   surprise-43-b13-ja        00 00 01 ½0 10 1½ ½0 ** 00 11 11 01 01 11 1½ ½½ 11 11 11 11  23.0/38  339.00
9   scidlet-261b2-ja          01 00 00 ½0 00 00 01 11 ** 10 11 01 10 11 0½ 11 10 11 11 11  21.0/38
10  gk-090-ja                 00 0½ 00 00 1½ 10 00 00 01 ** ½0 01 10 11 10 11 11 11 11 11  19.5/38
11  natwarlal-014-ja          00 1½ ½0 00 00 00 00 00 00 ½1 ** 01 1½ 10 11 11 11 11 11 11  19.0/38
12  knightcap-36-ja           10 ½0 10 ½0 00 10 10 10 10 10 10 ** 10 10 10 10 10 10 10 10  17.0/38  307.00
13  buzz-os-11-27-08-ja       00 00 00 00 10 00 01 10 01 01 0½ 01 ** 10 11 00 ½1 11 11 11  17.0/38  222.00
14  sjeng-112-ja              00 00 10 01 00 00 01 00 00 00 01 01 01 ** 11 01 11 01 11 11  16.0/38
15  mizar-30-ja               00 00 00 00 01 00 01 0½ 1½ 01 00 01 00 00 ** 11 11 11 1½ 11  15.5/38
16  zzzzzz-351-ja             00 00 ½0 00 00 00 ½0 ½½ 00 00 00 01 11 10 00 ** 10 ½1 11 11  12.5/38
17  mskcp-145-ja              00 00 00 00 00 00 00 00 01 00 00 01 ½0 00 00 01 ** ½1 11 11   9.0/38
18  exchess-054-beta-ja       00 00 00 00 00 00 00 00 00 00 00 01 00 10 00 ½0 ½0 ** 11 11   7.0/38
19  toledo-nanochess-2010-ja  00 00 00 00 00 00 00 00 00 00 00 01 00 00 0½ 00 00 00 ** ½½   2.5/38
20  Chess for Android         00 00 00 00 00 00 00 00 00 00 00 01 00 00 00 00 00 00 ½½ **   2.0/38

[Jim Ablett]

Thanks for the test Aart.
I can probably fix the GFC bug. I'll take a look.

Jim.

[Max]

Hi Aart,

looks like knightcap-36-ja lost _all_ games with the same color (perhaps black?) but scored 17 points. Could you please check the PGNs, if there went something wrong?

Thanks,
Max

[abik]

looks like knightcap-36-ja lost _all_ games with the same color (perhaps black?) but scored 17 points. Could you please check the PGNs, if there went something wrong?

Wow, I am very impressed you observed the problem from just the tournament table, Max!

Indeed, knightcap gave rise to one of the upcoming improvements I alluded to (specifically, it sends messages like "move num decreased 8 45", which were interpreted as illegal moves). Another problem with knightcap is that it fails to honor the fixed time per move (it consistently takes 10 seconds per move rather than the allowed 1 seconds per move for this tournament). So even though its defeats were somewhat unfair, so were its victories

[jdart]

Theoretically Arasan works under Android too (but I don't do official builds for that environment). I think Jim did a build of version 13.1.

--Jon

[abik]

Theoretically Arasan works under Android too (but I don't do official builds for that environment). I think Jim did a build of version 13.1.

--Jon

Hi Jon,
It would indeed be nice to add Arasan to the growing list of ARM-based Android engines. I got an Arasan binary a while back, but it never functioned correctly. I tested it again today, and although it prints a welcome message, it then becomes unresponsive to any input. I also copied the files kpk-b.bit and kpk-w.bit to the working directory, but that did not help either. Could you or Jim perhaps be so kind to send me a new binary for Android?

Code: Select all

./arasanx-123-ja
Arasan 12.3 Copyright 1994-2010 by Jon Dart

[Jim Ablett]

Theoretically Arasan works under Android too (but I don't do official builds for that environment). I think Jim did a build of version 13.1.

--Jon

Hi Jon,
It would indeed be nice to add Arasan to the growing list of ARM-based Android engines. I got an Arasan binary a while back, but it never functioned correctly. I tested it again today, and although it prints a welcome message, it then becomes unresponsive to any input. I also copied the files kpk-b.bit and kpk-w.bit to the working directory, but that did not help either. Could you or Jim perhaps be so kind to send me a new binary for Android?

Code: Select all

./arasanx-123-ja
Arasan 12.3 Copyright 1994-2010 by Jon Dart

Hi Aart/Jon,

I was going to talk to you about Arasan.
I can compile latest arasan src and the binary runs but moves instantly.
Also engine works straight away in xboard mode, but to get it to run in uci mode I must hack src so that uci protocol is default mode at startup/initialization.

I'll email you the binary so you can test it out.

Jim.

[jdart]

The latest version (13.4) has several UCI fixes, so if you plan to use UCI you should be on that version.

--Jon

[abik]

The latest version (13.4) has several UCI fixes, so if you plan to use UCI you should be on that version.

--Jon

Jim Ablett kindly compiled this version for me and it works well. Screenshots available at my blog.

[jdart]

Thanks, Jim! I plan to eventually support this platform myself but currently it's not part of my build/test process.

--Jon