Larson's Babies

govert · Post by **govert** » Tue Mar 16, 2010 3:06 pm

In order to learn a little about end game playing strength, I have carried out a tournament using Sune Larsson's End Game Test Suite

Intel(R) Core(TM)2 Duo CPU T9400 @ 2.53GHz with 3 064 MB Memory
Microsoft Windows XP Professional Service Pack 3 (Build 2600)

No tablebases were being used except Ivanhoe who used 3-4-5 RobboTripleBases
Rybka contempt=15

All engines were 32 bit-1CPU
Hash: 256 MB
Timecontrol: 30 sec + 2 sec

Results:

Code: Select all


Larsson
2010.03.08 - 2010.03.16
                                Score       1:Stockfish-163-32-ja 1CPU               2:Rybka 3 w32 1CPU                         
------------------------------------------------------------------------------------------------------------------------------
 1: Stockfish-163-32-ja 1CPU  122.5 / 200   XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0=01=11===1=1=1=01=10======0111==1===1==
 2: Rybka 3 w32 1CPU          111.0 / 200   1=10=00===0=0=0=10=01======1000==0===0== XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 
 3: IvanHoe999967             111.0 / 200   =11===011==00==00==010=0=0=01010=0=1===0 1==0=0=110==0==110=01===1==0100======00= 
 4: Komodo-10-32-ja            93.5 / 200   =0====00=0=10=00==101=1====0===0=0==1==0 00====00===101=1====101=1==0=0=====0==0= 
 5: Spark-0.3a-vc32            89.5 / 200   =0==0=10===010=00101010=0=01===00=0====1 0===0===0==0101=====0=0====00==001=====0 
 6: Crafty-23.2-win32          72.5 / 200   ==0=0===0=00=00==0010=0=010===010==0=000 00==0==00=0==0=1=0==01010=000010==0==0== 
------------------------------------------------------------------------------------------------------------------------------
 
   3:IvanHoe999967                          4:Komodo-10-32-ja                        
-------------------------------------------------------------------------------------
1: =00===100==11==11==101=1=1=10101=1=0===1 =1====11=1=01=11==010=0====1===1=1==0==1 
2: 0==1=1=001==1==001=10===0==1011======11= 11====11===010=0====010=0==1=1=====1==1= 
3: XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX ===1==1=0====01======10===1=11=0==1=11== 
4: ===0==0=1====10======01===0=00=1==0=00== XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 
5: =0=0=001=0==011001===0=01000=0=0=00====0 ==0==0=1==0101=110==1==0==1==0======1=0= 
6: 01====000=0==00110=101==0=000=001=00==0= ============011=01110=010=01=10=00=0==00 
-------------------------------------------------------------------------------------

   5:Spark-0.3a-vc32                        6:Crafty-23.2-win32                       
-------------------------------------------------------------------------------------
1: =1==1=01===101=11010101=1=10===11=1====0 ==1=1===1=11=11==1101=1=101===101==1=111
2: 1===1===1==1010=====1=1====11==110=====1 11==1==11=1==1=0=1==10101=111101==1==1==
3: =1=1=110=1==100110===1=10111=1=1=11====1 10====111=1==11001=010==1=111=110=11==1=
4: ==1==1=0==1010=001==0==1==0==1======0=1= ============100=10001=101=10=01=11=1==11
5: XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX ===1==10=1011=011=01=1=10=11=0=1==11==01
6: ===0==01=0100=100=10=0=01=00=1=0==00==10 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
-------------------------------------------------------------------------------------
600 games: +201 =307 -92

The Games

govert · Post by **govert** » Tue Mar 16, 2010 3:32 pm

If anyone is interested, I can publish the SCID generated statistics for each position like below:

Round 1-2:

[d] 8/1ppr1kp1/p1p4p/8/8/5P2/PPP1RKPP/8 w - - 0 1

A position which might have arisen from the Exchange Variation of the
Spanish. White's better pawn structure gives him some advantage, since he can
create a passed pawn on the kingside while Black, owing to his doubled pawns,
cannot do the same on the queenside. It is not clear if this advantage is
sufficient to win. In practical play I would expect wins and draws to be about
equally divided.

Code: Select all


Larsson
MASASI, 2010.03.08 - 2010.03.16
                                Score     Ry Iv St Cr Sp Ko
------------------------------------------------------------
 1: Rybka 3 w32 1CPU           7.5 / 10   XX 0= 1= 11 1= 11
 2: IvanHoe999967              6.5 / 10   1= XX =1 10 =1 ==
 3: Stockfish-163-32-ja 1CPU   5.0 / 10   0= =0 XX == =1 =1
 4: Crafty-23.2-win32          4.0 / 10   00 01 == XX == ==
 5: Spark-0.3a-vc32            3.5 / 10   0= =0 =0 == XX ==
 6: Komodo-10-32-ja            3.5 / 10   00 == =0 == == XX
------------------------------------------------------------
30 games: +9 =17 -4


1. Statistics and History

1.1 Statistics

                   Games     1-0     =-=     0-1    Score
-----------------------------------------------------------
 All report games     30       9      17       4    58.3%
-----------------------------------------------------------

1.2 Most frequent players (White)

  1:   5       2010  50%  2200  Crafty-23.2-win32 [7,9,16,19]
  2:   5       2010  40%  2200  Spark-0.3a-vc32 [11,12,16,20]
  3:   5       2010  50%  2200  Stockfish-163-32-ja 1CPU [3,10,18,21,23]
  4:   5       2010  90%  2200  Rybka 3 w32 1CPU [1,2,4,5,17]
  5:   5       2010  40%  2200  Komodo-10-32-ja [8,16,22,24]
  6:   5       2010  80%  2200  IvanHoe999967 [6,13,14,15]

1.3 Most frequent players (Black)

  1:   5       2010  30%  2200  Crafty-23.2-win32 [13,16,17,18]
  2:   5       2010  30%  2200  Spark-0.3a-vc32 [1,14,19,23,24]
  3:   5       2010  50%  2200  Stockfish-163-32-ja 1CPU [2,15,16,20,22]
  4:   5       2010  60%  2200  Rybka 3 w32 1CPU [7,11,13,21,22]
  5:   5       2010  30%  2200  Komodo-10-32-ja [5,6,9,10,12]
  6:   5       2010  50%  2200  IvanHoe999967 [3,4,7,8,11]

2. Result Trends

2.1 Result lengths and frequencies

                 Score      Game length             Frequency       
                          1-0    =-=    0-1    1-0     =-=     0-1  
 Report games    58.3%     47     43     61   30.0%   56.6%   13.3% 
 All games       59.0%     48     56     46   33.5%   51.1%   15.3% 

2.2 Shortest wins (White)

  1:  1-0(37) Stockfish-163-32-ja 1CPU 2200 - Komodo-10-32-ja 2200, MASASI 2010 [10]
  2:  1-0(37) Rybka 3 w32 1CPU 2200 - Spark-0.3a-vc32 2200, MASASI 2010 [1]
  3:  1-0(42) Rybka 3 w32 1CPU 2200 - Komodo-10-32-ja 2200, MASASI 2010 [5]
  4:  1-0(42) Rybka 3 w32 1CPU 2200 - Stockfish-163-32-ja 1CPU 2200, MASASI 2010 [2]
  5:  1-0(45) IvanHoe999967 2200 - Crafty-23.2-win32 2200, MASASI 2010 [13]

2.3 Shortest wins (Black)

  1:  0-1(55) Komodo-10-32-ja 2200 - Rybka 3 w32 1CPU 2200, MASASI 2010 [22]
  2:  0-1(60) Stockfish-163-32-ja 1CPU 2200 - IvanHoe999967 2200, MASASI 2010 [3]
  3:  0-1(63) Spark-0.3a-vc32 2200 - Stockfish-163-32-ja 1CPU 2200, MASASI 2010 [20]
  4:  0-1(68) Crafty-23.2-win32 2200 - Rybka 3 w32 1CPU 2200, MASASI 2010 [7]

3. Moves and Themes

3.1 Move orders reaching the report position

There was only one move order reaching this position:
  1:   (30)

3.2 Moves from the report position

    Move   ECO       Frequency    Score  AvElo Perf AvYear %Draws
 1: f4                29: 96.6%   56.8%  2200  2250  2010   59%
 2: Ke3                1:  3.3%  100.0%              2010    0%
_______________________________________________________________
TOTAL:                30:100.0%   58.3%  2200  2257  2010   57%

3.3 Endgames

Material at the end of each game:
                   P     BN      R   R,BN      Q   Q,BN    Q,R Q,R,BN
 Report games    13%     0%    47%     0%    33%     0%     7%     0%
 All games       16%    23%    14%     2%    25%    15%     5%     1%

4. Theory Table

-------------------------------------------------------------------------------
  +9 =17 -4 (17.5/30: 58%)
-------------------------------------------------------------------------------
       1        2        3        4        5        6        7        8     
-------------------------------------------------------------------------------
 1  f4[1]    g3[3]    b3[5]    fg5      Ke3      Rf2      Rf8      Kd3       6:
    Rd1[2]   Kf6[4]   g5       hg5      c5       Kg6      Re1      Re7[6]   67%

 2  ...      Kf3      g3[7]    g4[8]    Ke4      fg5      h3       Re3       4:
    ...      Kf6      c5       Rf1      g5       Kg5      Kh4      Rh1[9]   50%

 3  ...      ...      b3       c4[10]   g3       Kg4      gf4      h4        4:
    ...      ...      c5       g5[11]   Rf1      gf4      a5       a4[12]   63%

 4  ...      Ke3      b3       g3       Kf2      Kg2      c4       h3        4:
    Kf6      Rd1      Kf5[13]  Kg4      g6[14]   h5       h4       Kf5[15]  88%

 5  ...      Kf3      g3[16]   Re8[17]  Re2      Ke4      c3       Rc2       5:
    ...      g6       b6       Rd2      Rd7      Rd1      c5       Re1[18]  60%

 6  ...      ...      b3[19]   Ke3      Rd2      Kf3      Ke4      Kf3       3:
    ...      Rd1      c5[20]   b6       Re1      Rf1      Re1      Rf1[21]  33%

 7  ...      ...      g3       c4[23]   b3       c5       b4       Kg4       4:
    ...      ...      a5[22]   a4       a3       Rc1      Ra1      b5[24]   38%

-------------------------------------------------------------------------------


[1]  1.Ke3 g6 2.Rd2 Rd6 3.Rd4 c5 4.Rxd6 cxd6 5.Ke4 Ke6 6.c4 b5 7.b3 bxc4 8.bxc4 a5 9.a4 d5+ 1-0(37) Rybka 3 w32 1CPU 2200 - Spark-0.3a-vc32 2200, MASASI 2010.

[2]  1...g6 2.Ke3 Kf6 3.Rd2 Rd6 4.c4 Kf5 5.Rxd6 cxd6 6.h3 h5 7.g3 g5 8.fxg5 Kxg5 9.Ke4 Kf6 1-0(42) Rybka 3 w32 1CPU 2200 - Stockfish-163-32-ja 1CPU 2200, MASASI 2010.

[3]  2.Ke3 Kf6 3.Ke4 c5 4.c3 c4 5.h4 h5 6.Kf3 g6 7.Re4 b5 8.Re8 Rf1+ 9.Kg3 Rb1 0-1(60) Stockfish-163-32-ja 1CPU 2200 - IvanHoe999967 2200, MASASI 2010.

[4]  2...c5 3.Ke3 Kf6 4.b3 g6 5.Rd2 Rxd2 6.Kxd2 Kf5 7.Ke3 Kg4 8.Ke4 h5 9.Ke5 Kh3 =-=(31) Rybka 3 w32 1CPU 2200 - IvanHoe999967 2200, MASASI 2010.

[5]  3.Kf3 c5 4.b3 b6 5.Ke4 Rd4+ 6.Ke3 g5 7.Rf2 Kg6 8.c4 g4 9.Rd2 Rxd2 1-0(42) Rybka 3 w32 1CPU 2200 - Komodo-10-32-ja 2200, MASASI 2010.

[6]  9.a4 Rh7 =-=(36) IvanHoe999967 2200 - Komodo-10-32-ja 2200, MASASI 2010.

[7]  3.Ke4 c5 4.g4 g5 5.f5 h5 6.h3 Rh1 7.Re3 hxg4 8.hxg4 Rg1 9.Kf3 Rf1+ 1-0(50) Crafty-23.2-win32 - IvanHoe999967 2200, MASASI 2010;  4...c4 5.h4 Rh1 6.g5+ Kg6 7.f5+ Kf7 8.Ke5 Rxh4 9.g6+ Kf8 0-1(68) Crafty-23.2-win32 2200 - Rybka 3 w32 1CPU 2200, MASASI 2010.

[8]  4.a3 c4 5.Kg4 g6 6.h3 c6 7.Kf3 Rb1 8.c3 Rd1 9.Re8 Rf1+ =-=(36) Komodo-10-32-ja 2200 - IvanHoe999967 2200, MASASI 2010.

[9]  9.Rf3 b6 =-=(30) Crafty-23.2-win32 2200 - Komodo-10-32-ja 2200, MASASI 2010.

[10]  4.Ke4 Rd4+ 5.Ke3 Rd1 6.Rd2 Rxd2 7.Kxd2 Kf5 8.Ke3 g5 9.fxg5 hxg5 1-0(37) Stockfish-163-32-ja 1CPU 2200 - Komodo-10-32-ja 2200, MASASI 2010.

[11]  4...b6 5.h3 g6 6.g3 h5 7.g4 Rf1+ 8.Kg3 Rd1 9.gxh5 gxh5 =-=(39) Spark-0.3a-vc32 2200 - Rybka 3 w32 1CPU 2200, MASASI 2010.
       4...Rd3+ 5.Ke4 Rd1 6.h3 a5 7.g4 g6 8.Re3 Rd4+ 9.Kf3 a4 =-=(46) Spark-0.3a-vc32 2200 - IvanHoe999967 2200, MASASI 2010.

[12]  9.bxa4 h5+ =-=(31) Spark-0.3a-vc32 2200 - Komodo-10-32-ja 2200, MASASI 2010.

[13]  3...g6 4.Rd2 Re1+ 5.Kf3 Rf1+ 6.Kg3 Rc1 7.h3 b6 8.Kf3 Rf1+ 9.Ke3 Re1+ 1-0(45) IvanHoe999967 2200 - Crafty-23.2-win32 2200, MASASI 2010.
       3...c5 4.Rd2 Re1+ 5.Kf3 Rf1+ 6.Ke4 Ke6 7.c4 b6 8.Re2 g5 9.fxg5 hxg5 1-0(50) IvanHoe999967 2200 - Rybka 3 w32 1CPU 2200, MASASI 2010.

[14]  5...Ra1 6.a4 Kh3 7.Re7 Ra2 8.Rxg7 Rxc2+ 9.Kf3 Rc3+ 1-0(51) IvanHoe999967 2200 - Spark-0.3a-vc32 2200, MASASI 2010.

[15]  9.gxh4 Kxf4 =-=(101) IvanHoe999967 2200 - Stockfish-163-32-ja 1CPU 2200, MASASI 2010.

[16]  3.c3 Rd3+ 4.Ke4 Rd8 5.g4 Rd1 6.b3 a5 7.c4 Rg1 8.h3 Rh1 9.Re3 Rh2 =-=(26) Spark-0.3a-vc32 2200 - Crafty-23.2-win32 2200, MASASI 2010.
       3.b3 Rd1 4.c4 a5 5.Re8 Rf1+ 6.Kg3 Rc1 7.Rb8 Rc3+ 8.Kf2 Kf5 9.Rxb7 Rc2+ =-=(30) Komodo-10-32-ja 2200 - Crafty-23.2-win32 2200, MASASI 2010.
       3.a3 Rd1 4.Ke4 a5 5.g4 Ra1 6.Re3 a4 7.Rh3 Re1+ 8.Kd3 Rd1+ 9.Kc3 Rf1 =-=(30) Crafty-23.2-win32 2200 - Stockfish-163-32-ja 1CPU 2200, MASASI 2010.

[17]  4.Ke4 Rd1 5.b3 Ra1 6.c3 Rd1 7.b4 h5 8.Rf2 Re1+ 9.Kd3 Kf5 1-0(72) Rybka 3 w32 1CPU 2200 - Crafty-23.2-win32 2200, MASASI 2010.

[18]  9.Kd5 Re6 =-=(48) Stockfish-163-32-ja 1CPU 2200 - Crafty-23.2-win32 2200, MASASI 2010.

[19]  3.Ke4 a5 4.b3 a4 5.bxa4 Rd5 6.c4 Ra5 7.Rb2 b6 8.g4 Rxa4 9.Rc2 Ke6 =-=(39) Crafty-23.2-win32 2200 - Spark-0.3a-vc32 2200, MASASI 2010.

[20]  3...Ra1 4.c3 Rc1 5.c4 a5 6.Ke3 a4 7.Kd4 Rd1+ 8.Kc5 axb3 9.axb3 Rd3 0-1(63) Spark-0.3a-vc32 2200 - Stockfish-163-32-ja 1CPU 2200, MASASI 2010.

[21]  9.Kg4 g6 =-=(89) Stockfish-163-32-ja 1CPU 2200 - Rybka 3 w32 1CPU 2200, MASASI 2010.

[22]  3...c5 4.h3 c4 5.Ke4 g6 6.a4 Rh1 7.h4 Ra1 8.b3 Ra2 9.bxc4 Rxa4 0-1(55) Komodo-10-32-ja 2200 - Rybka 3 w32 1CPU 2200, MASASI 2010.
       3...Ra1 4.a3 Rb1 5.b3 Ra1 6.a4 c5 7.Rd2 Ra2 8.Rd7 Rxc2 9.Rxc7 Rc3+ =-=(72) Komodo-10-32-ja 2200 - Stockfish-163-32-ja 1CPU 2200, MASASI 2010.

[23]  4.b3 a4 5.bxa4 Rd5 6.Re8 Rc5 7.Rb8 b5 8.c4 bxc4 9.Ke3 c3 =-=(30) Stockfish-163-32-ja 1CPU 2200 - Spark-0.3a-vc32 2200, MASASI 2010.

[24]  9.cxb6 cxb6 =-=(20) Komodo-10-32-ja 2200 - Spark-0.3a-vc32 2200, MASASI 2010.

govert · Post by **govert** » Tue Mar 16, 2010 3:36 pm

Furthermore, I'm willing to add another free engine to the tournament.
Any suggestions?

Carlos777 · Post by **Carlos777** » Tue Mar 16, 2010 4:05 pm

Hi Martin,

Interesting result. I knew Stockfish was very strong at endgames, but I did not think it could beat R3 and Ivanhoe.

Furthermore, I'm willing to add another free engine to the tournament.
Any suggestions?

My first option would be Firebird 1.2, but I guess Ivanhoe being tested is enough. What about adding Protector 1.3.3? Other option could be Twisted Logic 20100131.

Best,
Carlos

govert · Post by **govert** » Tue Mar 16, 2010 7:41 pm

Only room for one of the Ippo family for now. I might replace Ivanhoe with FB later on if I decide to do a longer time control tournament with the top 4 or something...

I did a quick qualifying gauntlet between Protector and Twisted Logic and Protector came out on top:

Protector-133-32-ja - Twisted Logic 100131 : 23,5/40 +18-11=11

So, Protector has the spot at the moment, unless someone suggests another candidate

Carlos777 · Post by **Carlos777** » Wed Mar 17, 2010 4:41 pm

It seems that Critter is also strong at endgames. I think Spike is good too.

govert · Post by **govert** » Fri Mar 19, 2010 12:11 pm

I tested Critter and Spike against Protector and Critter gave Protector a surprisingly good run for the money. However Protector came out on top.

Carlos777 · Post by **Carlos777** » Fri Mar 19, 2010 1:30 pm

govert wrote:I tested Critter and Spike against Protector and Critter gave Protector a surprisingly good run for the money. However Protector came out on top.

It seems it is going to be Protector then. Thinker would be a good opponent, but I am not sure if it accepts to play from specific positions.

govert · Post by **govert** » Tue Mar 23, 2010 1:47 pm

Carlos777 wrote:
govert wrote:I tested Critter and Spike against Protector and Critter gave Protector a surprisingly good run for the money. However Protector came out on top.
It seems it is going to be Protector then. Thinker would be a good opponent, but I am not sure if it accepts to play from specific positions.

It does, and Thinker 54C inert actually beat protector 1.3.3 with a very narrow margin 20.5 vs. 19.5.

However, I've decided to go with Protector anyway, since a PV is nice to have.

govert · Post by **govert** » Tue Mar 23, 2010 1:49 pm

Updated results with Protector 1.3.3
Protector uses 5 men TBs

Code: Select all

                              Score     
------------------------------------------
 1: Stockfish-163-32-ja      148.5 / 240  
 2: IvanHoe999967            135.0 / 240  
 3: Rybka 3 w32 1CPU         131.5 / 240  
 4: Protector-133 TB         115.0 / 240   
 5: Komodo-10-32-ja          114.0 / 240 
 6: Spark-0.3a-vc32          109.5 / 240 
 7: Crafty-23.2-win32         86.5 / 240

Larson's Babies

Larson's Babies

Re: Larson's Babies

Re: Larson's Babies

Re: Larson's Babies

Re: Larson's Babies

Re: Larson's Babies

Re: Larson's Babies

Re: Larson's Babies

Re: Larson's Babies

Re: Larson's Babies