komodo 1.3 beta 1 test available

[Houdini]

Does Houdini include all underpromotions in the main search as well as in the quies or do you have restrictions?

In the main search all the promotions are considered, in the quiescence only Queen promotions, and Knight promotions with check.

[Don]

Does Houdini include all underpromotions in the main search as well as in the quies or do you have restrictions?

In the main search all the promotions are considered, in the quiescence only Queen promotions, and Knight promotions with check.

Ok, it's a simple matter for me to add that back in and test it.

Don

[Damir]

Hi Don

Any chance of having Komodo parameters back? I miss those when I want to try and tune the engine.. It is kinda boring now when there is nothing to play with.

Many thanks for the engine.

Damir

[Don]

Hi Don

Any chance of having Komodo parameters back? I miss those when I want to try and tune the engine.. It is kinda boring now when there is nothing to play with.

Many thanks for the engine.

Damir

Some interfaces cannot handle more than a few parameters so I left them out. One of the most popular interfaces can only accept 1 screen full of options and the evaluation paramaters would be several screens. I think it's the Fritz interface if I am remembering this correctly. I don't know about others.

The file method I was using caused me a lot of trouble and it's a pain to support it. Let me think about how I might add this support back in the future.

[Damir]

Maybe you can try and create a separate folder for a personallity parameters, without affecting engine's playing strength ?
It is true that Fritz Gui has some limitations, and not all parameters are shown. Arena GUI on the other hand does not have that kind of a problem, and displays all parameters without any problems. I think Chessbase can learn a lot from the way Arena folks are making their GUI, in the best user friendly way as possible.

[stevenaaus]

Ha ha.. i'd never seen this engine before. Here it is totally beating-up RobboLito in an interesting game.

"Komodo32 1.3b1 JA" vs "RobboLito verzione 0.085d"
[D]2r2nkr/3n1p2/2q1pP2/pR1p2P1/P1pP2QP/3N2B1/2R5/6K1 w - - 0 43

But it appears to take longer than it's per-move time slice on tough moves.

Code: Select all

[Event "Scid-vs-PC Tournament"]
[Site "?"]
[Date "2010.12.22"]
[Round "2"]
[White "Komodo32 1.3b1 JA"]
[Black "RobboLito verzione 0.085d"]
[Result "1-0"]
[Movetime "3"]

1. e4 Nc6 2. d4 d5 3. e5 Bf5 4. Nf3 e6 5. Bd3 Nge7 6. O-O a6 7. c3 Bxd3 8. Qxd3 Nf5 9. b4 Be7 10. Nbd2 O-O 11. Nb3 Qc8 12. a4 b6 13. h3 h6 14. Bd2 Qb7 15. Bf4 Nb8 16. Nbd2 Nd7 17. Rab1 a5 18. Rfc1 c6 19. b5 Rac8 20. bxc6 Qxc6 21. Rb5 Rc7 22. Rc2 Rd8 23. Ne1 Qb7 24. g4 Nh4 25. Bg3 Ng6 26. f4 Nh4 27. Qe3 Ng6 28. Nef3 Ngf8 29. f5 Rcc8 30. c4 Bb4 31. f6 gxf6 32. exf6 Kh7 33. g5 Bxd2 34. Nxd2 Ng6 35. h4 Rg8 36. c5 Qc6 37. Nf3 h5 38. Ne1 bxc5 39. Qe2 Rh8 40. Qxh5+ Kg8 41. Qg4 Ngf8 42. Nd3 c4 43. h5 c3 44. Ne5 Nxe5 45. Bxe5 Qe8 46. Rxa5 Rh7 47. Rb5 Rh8 48. Bd6 Nd7 49. Rb7 Qd8 50. Be7 Ne5 51. dxe5 Rc4 52. Qg3 Qf8 53. Bxf8 Rd4 54. g6 Rd1+ 55. Kf2 Rd2+ 56. Rxd2 cxd2 57. gxf7+ Kxf8 58. Qg7# 1-0 

[stevenaaus]

Hmmm... the per-move time control is definitely broke on my box (linux 32bit fedora7).

Here it is losing. I havent done analysis, but from move 50, i'm not sure it plays the end game well, but i guess the game was lost.

Code: Select all

[Event "Scid-vs-PC Tournament"]
[Site "?"]
[Date "2010.12.22"]
[Round "4"]
[White "Komodo32 1.3b1 JA"]
[Black "RobboLito verzione 0.085d"]
[Result "0-1"]
[Movetime "3"]

1. e4 Nc6 2. d4 e6 3. Nf3 d5 4. e5 Nge7 5. c3 Nf5 6. Bd3 Be7 7. O-O O-O 8. Na3 Bxa3 9. bxa3 b6 10. Rb1 h6 11. g4 Nfe7 12. Kh1 a5 13. g5 h5 14. Qe2 Nf5 15. Ne1 h4 16. Qg4 Nce7 17. Ng2 Ba6 18. Bxa6 Rxa6 19. Nxh4 Nxh4 20. Qxh4 Nf5 21. Qh5 g6 22. Qd1 Kg7 23. Qd3 Ra8 24. a4 Rh8 25. Kg1 Rh4 26. f3 c5 27. dxc5 bxc5 28. Qe2 Qc7 29. Rf2 Rxa4 30. Rf1 Rb8 31. Rxb8 Qxb8 32. Rf2 Rc4 33. Bb2 a4 34. f4 a3 35. Bxa3 Rxc3 36. Qb2 Qxb2 37. Bxb2 Rd3 38. Kf1 c4 39. Ke2 Rh3 40. Bc1 c3 41. Kd1 d4 42. Kc2 Ne7 43. a4 Nd5 44. Kb3 d3 45. Kc4 Kf8 46. a5 Ke7 47. Bd2 cxd2 48. Rxd2 Kd7 49. Rxd3 Rxd3 50. Kxd3 Nxf4+ 51. Ke4 Nh3 52. a6 Kc6 53. Kf3 Nxg5+ 54. Ke3 Kb6 55. Kd4 Nf3+ 56. Ke4 Nxh2 57. a7 Kxa7 58. Kf4 Kb6 59. Kg3 Nf1+ 60. Kf4 Kc5 61. Kf3 Kd4 62. Ke2 Ng3+ 63. Kd2 Ne4+ 64. Kd1 Kxe5 65. Ke2 f5 66. Kf3 g5 67. Ke3 g4 68. Kd3 g3 69. Ke3 g2 70. Kd3 g1=Q 71. Kc4 Qb6 72. Kd3 Qc5 73. Ke2 Qd4 74. Ke1 Qd2+ 75. Kf1 Qf2# 0-1

[Don]

I'm hoping to make the official release soon. Are there any further bugs to report?

I think we have resolved an issue that causes the 64 bit version it to occasionally crash on some systems when auto-testing.

Komodo 1.3 beta 1 test by Don Dailey & Larry Kaufman

Windows/Linux 64/32 Intel p.g.o

This is an unoffical beta test version.

What's new
Improved evaluation.
Improved search.
30 ELO stronger. (conservative estimate)


Download:
http://www.mediafire.com/?cb3do6a9i69sqag
Mirror:
http://good.net/_C6znfW9ms

Jim.

[beachknight]

First tests show +55 improvement over 1.2

http://www.talkchess.com/forum/viewtopic.php?t=37287

Congrats, Komodo team.

[Don]

Why is it not running from the command line?

After starting it displays its name but accepts no input.

This was similar with 1.2 but it accepted 'uci' although only to display another line, uci_K = 0x00002b5f0.

I do not understand windows I/O - I had to ask for Jim's help in figuring this out. Even though it did not work from a windows dos prompt it still worked on GUI's. Anyway, I believe we have fixed this in the upcoming release.

Don