A privat RobboLito 0.21Q test under CEGT-Blitz conditions

[Tomcass]

I repeated the test. After 240 additional games,

Testing Robbolito 0.20Q = 120 games
I7 980 3.33 Ghz.
6 real cores
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

201209Robbolito 0.20Qi_2 2012

RobboLito 0.20Q x64 SSE4.2_6 - Critter 1.6 64-bitx6_nob 18.0 - 22.0 +3/=30/-7 45.00%
RobboLito 0.20Q x64 SSE4.2_6 - Deep Rybka 4.1 SSE42 x64 (x6) 18.0 - 22.0 +7/=22/-11 45.00%
RobboLito 0.20Q x64 SSE4.2_6 - Houdini 2.0c Pro x64(x6) 12.0 - 28.0 +3/=18/-19 30.00%

Average Score : 40.00%.

Games: http://www.mediafire.com/?1i3b54e3t73cw6j

Testing Robbolito 0.21Q = 120 games

I7 980 3.33 Ghz.
6 real cores
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

201209Robbolito 0.21Q_2

RobboLito 0.21Q x64 SSE4.2_ -Critter 1.6 64-bitx6_nob 18.5 - 21.5 +4/=29/-7 46.25%
RobboLito 0.21Q x64 SSE4.2_ -Deep Rybka 4.1 SSE42 x64 (x6) 17.0 - 23.0 +6/=22/-12 42.50%
RobboLito 0.21Q x64 SSE4.2_ -Houdini 2.0c Pro x64(x6) 15.0 - 25.0 +6/=18/-16 37.50%

Average score: 42.08%.

Games: http://www.mediafire.com/?6lyd45kn22nae79

So that in my tests, after 480 games under the same conditions, Robbolito 0.21Q is stronger than Robbolito 0.20Q. 42.71% against 40.21%.

Regards from Barcelona.

Tom.

[Tomcass]

Testing Robbolito 0.9.02Qi =120 games

I7 980 3.33 Ghz.
1 core
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 minutes +2 sec/ game.
Ponder: On
No tablebases

201210Robbolito 0.9.02Q

Robbo 09.02Qi x64 SSE4.2LP- Deep Rybka 4.1 SSE42 x64(x1) 19.5 - 20.5 +11/=17/-12 48.75%
Robbo 09.02Qi x64 SSE4.2LP - Houdini 2.0c Pro x64(x1) 15.5 - 24.5 +4/=23/-13 38.75%
Robbo 09.02Qi x64 SSE4.2LP - Komodo 5 64-bit_x1 19.5 - 20.5 +10/=19/-11 48.75%

Average score: 45.42%

Games: http://www.mediafire.com/?9ouxce3kpfd93my

This result is substantially better than the one scored by R0.20Q and R0.21Q versions. But please note that only one core has been used. Perhaps the scalability from one to multi-core is an area of potential improvement for this exciting engine.

Regards,

Tom.

[kranium]

http://www.chess2u.com/t6341p15-robboli ... 4150-games
RobboLito 0.20Q x64 gauntlet / RobboLito 0.21Q x64 gauntlet

Code: Select all

    Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 Houdini 2.0c Pro x64           : 3335   17  17   811    63.7 %   3237   49.0 %
  2 Critter 1.6a 64-bit            : 3277   15  15   811    55.7 %   3237   61.3 %
  3 RobboLito 0.21Q x64            : 3238    9   9  2499    46.5 %   3262   56.5 %
  4 RobboLito 0.20Q x64            : 3237   16  16   750    46.4 %   3262   58.4 %
  5 IvanHoe 9.46h x64              : 3229   13  13   815    48.8 %   3237   69.3 %
  6 Stockfish 2.3.1 JA 64-bit      : 3209   17  17   812    46.0 %   3237   48.2 %

ELOstat - Start Elo = 3250

[ThatsIt]

[...snip...]

Code: Select all

    Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 Houdini 2.0c Pro x64           : 3335   17  17   811    63.7 %   3237   49.0 %
  2 Critter 1.6a 64-bit            : 3277   15  15   811    55.7 %   3237   61.3 %
  3 RobboLito 0.21Q x64            : 3238    9   9  2499    46.5 %   3262   56.5 %
  4 RobboLito 0.20Q x64            : 3237   16  16   750    46.4 %   3262   58.4 %
  5 IvanHoe 9.46h x64              : 3229   13  13   815    48.8 %   3237   69.3 %
  6 Stockfish 2.3.1 JA 64-bit      : 3209   17  17   812    46.0 %   3237   48.2 %

ELOstat - Start Elo = 3250

Nearly 100 points behind Houdini !
Thats worser than my results.

[kranium]

[...snip...]

Code: Select all

    Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 Houdini 2.0c Pro x64           : 3335   17  17   811    63.7 %   3237   49.0 %
  2 Critter 1.6a 64-bit            : 3277   15  15   811    55.7 %   3237   61.3 %
  3 RobboLito 0.21Q x64            : 3238    9   9  2499    46.5 %   3262   56.5 %
  4 RobboLito 0.20Q x64            : 3237   16  16   750    46.4 %   3262   58.4 %
  5 IvanHoe 9.46h x64              : 3229   13  13   815    48.8 %   3237   69.3 %
  6 Stockfish 2.3.1 JA 64-bit      : 3209   17  17   812    46.0 %   3237   48.2 %

ELOstat - Start Elo = 3250

Nearly 100 points behind Houdini !
Thats worser than my results.

yes, w/ Critter 58 behind, and Stockfish 124 behind
and it will soon get even worse