Komodo 3 release

[belkov]

[d]3rk3/1R3p2/p7/3N2p1/4P1nr/R2B1p2/P1P3bP/6K1 b - - 0 26

Here both Komodo 2.03 and Komodo 3 (in single thread mode) give ...Rxh2 as a mate in 2 which is totally wrong because white has Nf6+, winning for white.

If I put it on 2 or 3 threads it gives me random results: sometimes it sees ...Rxd5! and sometimes it doesn't. It appears as if the engine just stops calculating when it sees a mate irrespective if the mate can be evaded.

[Don]

[d]3rk3/1R3p2/p7/3N2p1/4P1nr/R2B1p2/P1P3bP/6K1 b - - 0 26

Here both Komodo 2.03 and Komodo 3 (in single thread mode) give ...Rxh2 as a mate in 2 which is totally wrong because white has Nf6+, winning for white.

If I put it on 2 or 3 threads it gives me random results: sometimes it sees ...Rxd5! and sometimes it doesn't. It appears as if the engine just stops calculating when it sees a mate irrespective if the mate can be evaded.

I will take a look at this and fix this bug. Thanks for the report.

Don

[EroSennin]

Congratulations,Komodo is No.2 behind Houdini on the IPON single core engines.It seems the MP version should be a killer!Good Luck and good work!

I see no komodo 3 in the ipon rating list.

[Don]

Congratulations,Komodo is No.2 behind Houdini on the IPON single core engines.It seems the MP version should be a killer!Good Luck and good work!

I see no komodo 3 in the ipon rating list.

I had to press control f5 to get the page to update once. For some reason I've always had trouble with it. But if you do that and go the ratings tab you should see it.

Here is the table you should see:

Code: Select all

     Name                      Elo    +    -   games scor oppo.  draws
 
   1 Houdini 1.5a             3010   11   11  4000   79%  2777   26% 
   2 Komodo64 3 SSE42         2966   14   14  2100   76%  2770   29% 
   3 Deep Rybka 4.1 SSE42     2955   11   11  3000   73%  2786   36% 
   4 Critter 1.2              2954   12   12  2600   74%  2777   36% 
   5 Stockfish 2.1.1 JA       2943   12   12  2900   72%  2781   35% 
   6 Naum 4.2                 2826    8    8  6200   59%  2764   40% 
   7 Deep Shredder 12         2800    7    7  7300   56%  2757   38% 
   8 Gull 1.2                 2797   11   10  3200   51%  2791   36% 
   9 Deep Sjeng c't 2010 32b  2787    9    9  4200   50%  2788   39% 
  10 Spike 1.4 32b            2783   10   10  3300   49%  2791   39% 
  11 Deep Fritz 12 32b        2780    8    8  6100   53%  2760   38% 
  12 Protector 1.4.0 x64      2760   11   10  3300   47%  2780   37% 
  13 Hannibal 1.1             2758   11   11  2700   46%  2790   40% 
     spark-1.0 SSE42          2758   10   10  3900   45%  2790   40% 
  15 HIARCS 13.2 MP 32b       2752   10   10  3700   45%  2787   37% 
  16 Deep Junior 12.5         2730   11   11  3000   42%  2793   34% 
  17 Zappa Mexico II          2716    7    7  8600   46%  2744   37% 
  18 Deep Onno 1-2-70         2686    8    9  5200   38%  2775   37% 
  19 Strelka 2.0 B            2674   11   11  3300   34%  2792   34% 
  20 Toga II 1.4 beta5c BB    2673    7    7  8100   40%  2745   38% 
  21 Loop 13.6/2007           2620    8    8  7100   32%  2753   34% 
  22 Jonny 4.00 32b           2616   10   10  4400   28%  2787   29% 
  23 Crafty 23.3 JA           2598   10   10  4300   26%  2783   28% 

[JohnS]

[d]3rk3/1R3p2/p7/3N2p1/4P1nr/R2B1p2/P1P3bP/6K1 b - - 0 26

Here both Komodo 2.03 and Komodo 3 (in single thread mode) give ...Rxh2 as a mate in 2 which is totally wrong because white has Nf6+, winning for white.

If I put it on 2 or 3 threads it gives me random results: sometimes it sees ...Rxd5! and sometimes it doesn't. It appears as if the engine just stops calculating when it sees a mate irrespective if the mate can be evaded.

I will take a look at this and fix this bug. Thanks for the report.

Don

There may be a probelm in multi-pv mode when in Chessbase as well.

In Chessbase for 2 and 3 variations it gives

1. -/+ (-1.18): 1...Kf8 2.Bxa6 Rxh2
2. +- (#1995): 1...Rxh2 2.Re7+ Kf8 3.Bxa6 f2#

and

1. +- (#1990): 2.Bb5+ Kf8 3.Rxf3 Bxf3 4.Rxf7+ Kxf7 5.Be8+ Rxe8 6.c4 Rh1#
2. +- (#1994): 2.Nc7+ Kf8 3.Ne6+ Kg8 4.Nxg5 f2#
3. +/- (0.91): 2.Nf6+ Nxf6 3.Kxh2 Kf8 4.Ra5 g4 5.Kg3 Rd4 6.Rf5 Kg7 7.e5 Nd7 8.Rg5+ Kh6 9.Rxg4 Rxg4+ 10.Kxg4 Nxe5+ 11.Kg3 a5 12.Rb6+ Kg7 13.Rb5 Nxd3 14.cxd3

[Dave_N]

There is a problem in Komodo 3 with some empassant captures, for example I was looking at this game

r3r1k1/pp4pp/1bn1b3/4Pp2/4QB1q/2N5/PP1N1PPP/3R1RK1 w - f6 0 17

obviously the revealed attack from the Bishop would give black some play, however Komodo seems to stop evaluating. My latest result from Fritz 12 was the move Xc1 ... possibly a Fritz problem. I appreciate the move might be completely bad.

[Dave_N]

If I play the move then copy and paste the new FEN

r3r1k1/pp4pp/1bn1bP2/8/4QB1q/2N5/PP1N1PPP/3R1RK1 b - - 0 1

then Komodo can evaluate ... so perhaps its the empassant string in the FEN notation.