bright-0.2c released

[Edward German]

I decided to release bright 0.2c!
It is publicly available from

http://brightchess.googlepages.com

Merry Xmas,

-
Allard

Thanks!

Here my first test and my first game (with german comment):

Mein Dank an den Programmierer!

Die UCI-Engine ist MP-fähig und kann sogar die Scorpio-Bitbases benützen.

Bei mir nutzt bright-0.2c im Mittelspiel auf 2 Kernen mit je 2.66 GHz etwa 3000 kn/s.

Stellung zur Erkundung der Max. Suchtiefe:
[d]2b3k1/1p1p1p1p/1P1P1P1P/8/8/p1p1p1p1/P1P1P1P1/2K2B2 w - - 0

Analysis by bright-0.2c:

1.Kd1
= (-0.04) Tiefe: 1/1 00:00:00
1.Kd1 Kf8
= (0.04) Tiefe: 2/2 00:00:00
1.Kd1 Kf8 2.Ke1
= (-0.04) Tiefe: 3/3 00:00:00
1.Kd1 Kf8 2.Ke1 Kg8
= (0.04) Tiefe: 4/4 00:00:00
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 5/5 00:00:00
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 6/6 00:00:00 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 7/6 00:00:00 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 8/6 00:00:00 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 9/8 00:00:01 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 10/8 00:00:01 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 11/8 00:00:01 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 12/8 00:00:01 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 13/8 00:00:01 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 66/11 00:00:01 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 78/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 84/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 86/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 88/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 90/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 92/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 94/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 96/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 98/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 100/11 00:00:02 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 101/11 00:00:03 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 104/11 00:00:03 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 105/11 00:00:03 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 108/11 00:00:03 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 110/11 00:00:03 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 112/11 00:00:03 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 114/11 00:00:03 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 116/11 00:00:04 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 118/11 00:00:04 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 120/11 00:00:04 2691mN
1.Kd1 Kf8 2.Ke1 Kg8 3.Kd1
= (0.00) Tiefe: 120/11 00:00:04 2691mN

(Nemeth, Privat 25.12.2007)

Die ersten "offline" tests sehen sehr gut aus!

Bei meiner bisher ersten und einzigen Partie auf Schach.de, konnte bright-0.2c auf einem Core 2 Duo 2.66 GHz gegen Rybka auf Quad ein Remis holen.

Bright-0.2c ist auf dem besten Weg!

English: Bright-0.2c is on the best road!

PGN:

[Event "Wertungspartie, 3m + 0s"]
[Site "Maschinenraum"]
[Date "2007.12.25"]
[Round "?"]
[White "Solist, bright-0.2c"]
[Black "Valheru, Rybka 2.3.2a mp"]
[Result "1/2-1/2"]
[ECO "B92"]
[WhiteElo "2601"]
[BlackElo "2643"]
[Annotator "0.11;0.22"]
[PlyCount "142"]
[EventDate "2007.12.25"]
[TimeControl "180"]

{Rybka 2.3.2a mp: 18.3 ply; 349kN/s AMD Opteron(tm) Processor 854 2800MHz, (4
threads)} 1. e4 {B/0 0} c5 {0} 2. Nf3 {B/0 0} d6 {0} 3. d4 {B/0 0} cxd4 {0} 4.
Nxd4 {B/0 0} Nf6 {0} 5. Nc3 {B/0 0} a6 {0} 6. Be2 {B/0 0} e5 {0} 7. Nb3 {B/0 0}
Be7 {0} 8. O-O {B/0 0} O-O {0} 9. Be3 {B/0 0} Be6 {0} 10. Qd2 {B/0 0} Nbd7 {0}
11. a4 {B/0 0} Rc8 {0} 12. a5 {B/0 0} Qc7 {0} 13. Rfd1 {B/0 0} Qc6 {0} 14. Bf3
{B/0 0} Rfe8 {0} 15. Qe1 {B/0 0} h6 {0} 16. h3 {B/0 0} Qc7 {0} 17. Nc1 {B/0 0}
Bf8 {0} 18. N1e2 {B/0 0} Qc4 {0.22/14 12} 19. Ra4 {0.11/13 9} Qc7 {0.21/17 14}
20. Ng3 {0.14/15 0} Red8 {0.19/16 23} 21. Rd2 {0.21/14 10} b5 {0.25/16 12} 22.
axb6 {0.43/15 8} Nxb6 {0.17/17 4} 23. Bxb6 {0.38/15 3} Qxb6 {0.18/18 15} 24.
Qa1 {0.31/16 0} Qb7 {0.21/16 9} 25. Be2 {0.41/13 6} Ra8 {0.24/16 6} 26. Rd1 {
(Bxa6) 0.31/11 5} a5 {0.18/12 2} 27. Nh5 {0.27/11 4} Nd7 {0.19/13 1} 28. Rxa5 {
0.27/12 4} g6 {0.21/15 1} 29. Ng3 {0.54/13 4} h5 {0.23/15 3} 30. Nb5 {
(b4) 0.58/12 10} Rdb8 {0.04/12 2} 31. Rxa8 {(Nxd6) 0.68/12 4} Rxa8 {0.03/10 0}
32. Qc1 {0.58/13 3} Qb6 {0.25/12 3} 33. Qg5 {(c4) 0.69/12 4} Kh7 {0.24/11 1}
34. c4 {0.69/11 3} Ra2 {0.24/13 1} 35. Qc1 {(Nc3) 0.70/12 3} Nc5 {0.21/12 1}
36. Qc2 {0.69/11 1} Kg7 {0.16/13 1} 37. Bf1 {(Nf1) 0.60/12 10} Ra4 {-0.05/12 2}
38. Kh2 {(h4) 0.36/12 8} h4 {-0.10/11 1} 39. Nh1 {0.35/13 1} Qa5 {-0.15/11 0}
40. f3 {(Nxd6) 0.53/12 2} Qb4 {0.00/11 2} 41. Nxd6 {0.63/13 1} Bxd6 {0.00/13 1}
42. Rxd6 {0.52/14 1} Nb3 {0.07/15 1} 43. Rd1 {0.44/14 2} Ra2 {0.09/17 3} 44.
Qb1 {(Nf2) 0.47/14 2} Ra1 {0.14/15 2} 45. Qd3 {0.21/16 0} Bxc4 {0.24/15 1} 46.
Qxc4 {0.21/17 1} Qxc4 {0.24/21 0} 47. Bxc4 {0.29/20 1} Rxd1 {0.25/22 1} 48.
Bxb3 {0.27/20 1} Rb1 {0.26/22 0} 49. Bc4 {0.33/19 1} Rxb2 {0.26/22 0} 50. Kg1 {
0.26/20 2} Rc2 {0.26/21 0} 51. Bd3 {0.41/18 1} Rc1+ {0.26/22 1} 52. Kh2 {
0.40/20 0} Kf6 {0.26/22 1} 53. Nf2 {0.42/19 1} Kg5 {0.26/21 7} 54. Bb5 {
(Ng4) 0.46/17 2} Rc5 {0.26/20 1} 55. Be8 {0.43/17 0} f5 {0.26/20 0} 56. Nd3 {
0.50/17 1} Rc8 {0.29/20 1} 57. Bd7 {0.51/16 0} Rc7 {0.29/20 2} 58. Bb5 {
0.52/17 0} Kf6 {0.32/19 1} 59. Nb4 {(Kg1) 0.36/15 0} Rb7 {0.25/17 1} 60. Nd5+ {
0.39/15 2} Kg5 {0.25/21 0} 61. Ba6 {(Bc6) 0.39/16 1} Rb1 {0.22/16 1} 62. Bd3 {
(Bc4) 0.31/15 2} Rd1 {0.08/19 1} 63. Ba6 {(Bb5) 0.14/16 1} Rb1 {0.01/20 1} 64.
Nc7 {(Bc4) 0.27/16 2} Rc1 {0.18/16 1} 65. Ne6+ {(Ne8) 0.25/17 2} Kf6 {0.16/16 0
} 66. Nf8 {0.19/17 1} Rc7 {0.19/18 0} 67. Bb5 {0.14/17 1} Rc1 {0.21/19 2} 68.
Nd7+ {0.00/16 0} Ke6 {0.21/19 0} 69. Nf8+ {(Nb6) 0.00/16 1} Kf6 {0.01/22 1} 70.
Ba6 {(Nd7+) 0.00/17 0} Rc7 {0.01/19 0} 71. Bb5 {0.00/17 1} Rc1 {
(Lag: Av=0.42s, max=1.1s) 0.01/23 1} 1/2-1/2

Klasse gespielt von bright-0.2c! Weiter so!

[Eraserheads]

"Yes, one of the best free engines now, and certainly the best free "

Hi,

Are there tests that prove Bright's tactical superioty over other free engines? What about Pro Deo's Q3 Personality? Or Little Goliath? How does Bright fare against them in tactical suites?

[gerold]

"Yes, one of the best free engines now, and certainly the best free "

Hi,

Are there tests that prove Bright's tactical superioty over other free engines? What about Pro Deo's Q3 Personality? Or Little Goliath? How does Bright fare against them in tactical suites?

Try the new Arasan8 test. Bright will beat Pro Deo in canter. Little Goliath is even more weaker, its tactical superiority, unfortunately, is only a legend by now.

been running games Bright v Rybka 1.1 beta. It does very well
in fast time controls.

[Dann Corbit]

I decided to release bright 0.2c!
It is publicly available from

http://brightchess.googlepages.com

Merry Xmas,

-
Allard

At first, I thought Thinker had met his match for Elo/byte, but I see you have packed the executable with UPX, so the real size is 386,048 bytes.

Nevertheless, it is a remarkably strong program even for this size.

[ChessAddict]

Is this a multi-processor engine?

[ChessAddict]

Is this a multi-processor engine?

It is.

Thanks. I'll try it out and see how it works.

[Ovyron]

Q3 surprises me. Nevertheless it may be much weaker than the default in match play.

It is. It's because Q3 wastes too much time searching for something that is not there. But when there's something, it's better at finding it than default.

[ernst]

Thanks Allard for this great engine.

However I am unable to use more than 1450MB hash or the memory used drops again. I have a Q6600 and 4GB RAM.

Futhermore when you look at the output from the engine, there are two anomalies.

Code: Select all

### from bright-0.2c (1): info currmove e2e4 currmovenumber 1 time 110406 nodes 561315392 nps 5084 hashfull 2026

Both the nps and the hashfull info aren't conform the uci standard.

[Allard Siemelink]

Hi Ernst,

The nps in the uci output is actually knps (nps/1024), does it cause you any trouble?

You probaby discovered a bug in the hash allocation, I will look into that.
Meanwhile, you may try specifying an exact power of two of for the hash size, e.g. 1024 or 2048 Mb.

Thanks Allard for this great engine.

However I am unable to use more than 1450MB hash or the memory used drops again. I have a Q6600 and 4GB RAM.

Futhermore when you look at the output from the engine, there are two anomalies.

Code: Select all

### from bright-0.2c (1): info currmove e2e4 currmovenumber 1 time 110406 nodes 561315392 nps 5084 hashfull 2026

Both the nps and the hashfull info aren't conform the uci standard.

[Allard Siemelink]

Yes, it is compressed with upx; I like my exe's small

At first, I thought Thinker had met his match for Elo/byte, but I see you have packed the executable with UPX, so the real size is 386,048 bytes.

Nevertheless, it is a remarkably strong program even for this size.