Not logged inCSS-Forum
Forum CSS-Online Help Search Login
CSS-Shop Impressum Datenschutz
Up Topic Hauptforen / CSS-Forum / FireBird 1.2 against Rybka 3
- - By Orlando de la Vega Date 2010-03-15 15:28 Edited 2010-03-15 15:37
FireBird 1.2 w32 against Rybka 3 32-bit

PlaceEngineAuthorCountryPointsSingle results
1FireBird 1.2 w32Kranium & Sentinel31.0/50½0½½1½½011-11½1001½½½-1½11110½½1-1½110½101½-0011½1½1½0
2Rybka 3 32-bitV. Rajlich & L. Kaufman19.0/50½1½½0½½100-00½0110½½½-0½00001½½0-0½001½010½-1100½0½0½1




50 of 50 matches played.
FireBird 1.2 w32 v/s Rybka 3 32-bit (50: + 22,= 18,- 10)

Start of tournament: ......... 2010.01.15, 10:14:25
Town/ Country: .............. Dax, France
Level: ............................ Blitz 40/1' repeated
Hardware: ...................... AMD Phenom(TM) 9750 Quad-Core
Engines: ......................... FireBird 1.2 w32 / Rybka 3 32-bit both using 4 threads
Hash: ............................. 256 MB / 256 MB
GUI: ............................... Shredder Classic 3
OS: ................................ Windows Seven Ultimate
Ponder OFF
Nalimov TB ..................... 3+4+5
Book: ............................. Shredder book, stopped using after 5 moves.
Openings: ....................... 25 different openings randomly chosen by GUI, large mode.

FireBird is a better endgame opponent than Rybka. For example this position, Rybka 3 see victory for whites, but FireBird see draw!



Best regards,
Orlando
Parent - By Bram Mourik Date 2010-03-17 21:05
Hi Orlando,

Bei mir gab es mit Zeitbedingung 4m 2s, auch meistens so etwa 60% besser fur FB 1.0 beta - 1.1 und 1.2

Hast du auch Tests zwischen Firebird und Rybka mit längere Zeit laufen lassen ?

zB 25 minuten pro Partie ?

grusse Bram
Parent - - By Ernest Bonnem Date 2010-03-18 01:19
[quote="Orlando de la Vega"]For example this position, Rybka 3 see victory for whites, [/quote]
Wieder mal falsch!
8/3k1p2/8/7p/2p5/P1P5/1P2KP2/8 w - - 0 1
Analysis by Rybka 3 32-bit:
1.Ke3 Kc6 2.Kf3
  +-  (3.77)   Depth: 15   00:00:01  204kN
1.Ke3 Kc6[] 2.Kf4
  +/=  (0.38)   Depth: 16   00:00:12  1869kN, tb=50
1.Ke3 Kc6[] 2.Kf4 Kb5[] 3.Ke3
  +/=  (0.41)   Depth: 17   00:00:15  2281kN, tb=111
Parent - - By Orlando de la Vega Date 2010-03-18 09:58
Hi Ernest,

can you tell me how much Hash do you allocate to Rybka here ?

Best,
Orlando
Parent - - By Ernest Bonnem Date 2010-03-18 19:42
Hash=512 MB,  Nalimov3-4-5
2-cpu
Here is another try:
8/3k1p2/8/7p/2p5/P1P5/1P2KP2/8 w - - 0 1
Analysis by Rybka 3 32-bit:
1.Kf3 Kc6 2.Kf4
  +-  (4.47)   Depth: 15   00:00:01  132kN
1.Kf3 Kc6 2.Kf4 Kb5 3.Kg5 Ka4[] 4.Kxh5 Kb3[] 5.Kg4 Kxb2[] 6.a4[] Kxc3[] 7.a5[] Kb2 8.a6 c3 9.a7 c2 10.a8Q c1Q 11.Qb7+ Ka3 12.Qf3+ Kb4
  +/=  (0.32)   Depth: 16   00:00:07  1151kN, tb=33
1.Kf3 Kc6 2.Kf4 Kb5 3.Kg5 Ka4[] 4.Kxh5 Kb3[] 5.Kg4 Kxb2[] 6.a4[] Kxc3[] 7.a5[] Kb2 8.a6 c3 9.a7 c2 10.a8Q c1Q 11.Qb7+ Ka3 12.Qf3+ Kb2 13.Qf6+
  +/=  (0.32)   Depth: 17   00:00:09  1520kN, tb=59
1.Kf3 Kc6 2.Kf4 Kb5
  +/=  (0.32)   Depth: 18   00:00:13  2052kN, tb=83
1.Kf3 Kc6[] 2.Kf4 Kb5
  +/=  (0.36)   Depth: 19   00:00:20  2996kN, tb=321

But with 1-cpu, it takes a bit longer...
Parent - - By Orlando de la Vega Date 2010-03-18 21:12 Edited 2010-03-18 21:15
Hi Ernest,

your results are very strange because you obtained two different analysis.

[quote="Ernest Bonnem"]
8/3k1p2/8/7p/2p5/P1P5/1P2KP2/8 w - - 0 1
Analysis by Rybka 3 32-bit:
1.Ke3 Kc6 2.Kf3
  +-  (3.77)   Depth: 15   00:00:01  204kN
1.Ke3 Kc6[] 2.Kf4
  +/=  (0.38)   Depth: 16   00:00:12  1869kN, tb=50
1.Ke3 Kc6[] 2.Kf4 Kb5[] 3.Ke3
  +/=  (0.41)   Depth: 17   00:00:15  2281kN, tb=111
[/quote]

[quote="Ernest Bonnem"]
Hash=512 MB,  Nalimov3-4-5
2-cpu
Here is another try:
8/3k1p2/8/7p/2p5/P1P5/1P2KP2/8 w - - 0 1
Analysis by Rybka 3 32-bit:
1.Kf3 Kc6 2.Kf4
  +-  (4.47)   Depth: 15   00:00:01  132kN
1.Kf3 Kc6 2.Kf4 Kb5 3.Kg5 Ka4[] 4.Kxh5 Kb3[] 5.Kg4 Kxb2[] 6.a4[] Kxc3[] 7.a5[] Kb2 8.a6 c3 9.a7 c2 10.a8Q c1Q 11.Qb7+ Ka3 12.Qf3+ Kb4
  +/=  (0.32)   Depth: 16   00:00:07  1151kN, tb=33
1.Kf3 Kc6 2.Kf4 Kb5 3.Kg5 Ka4[] 4.Kxh5 Kb3[] 5.Kg4 Kxb2[] 6.a4[] Kxc3[] 7.a5[] Kb2 8.a6 c3 9.a7 c2 10.a8Q c1Q 11.Qb7+ Ka3 12.Qf3+ Kb2 13.Qf6+
  +/=  (0.32)   Depth: 17   00:00:09  1520kN, tb=59
1.Kf3 Kc6 2.Kf4 Kb5
  +/=  (0.32)   Depth: 18   00:00:13  2052kN, tb=83
1.Kf3 Kc6[] 2.Kf4 Kb5
  +/=  (0.36)   Depth: 19   00:00:20  2996kN, tb=321
[/quote]

Are you really sure that you did these tests after a Clear Hash ?

Let me show you my analysis of Rybka 3 1-cpu with 512 mb of Hash and Nalimov TB 3+4+5:

Code:
  2.00   0:00   +1.75   1.Kf3 (36) 2
  3.00   0:00   +1.95++  1.Kf3 (95) 2
  3.00   0:00   +2.15++  1.Kf3 (123) 3
  3.00   0:00   +2.15   1.Kf3 (135) 4
  4.00   0:00   +2.35++  1.Kf3 (185) 3
  4.00   0:00   +2.46   1.Kf3 (200) 4
  5.00   0:00   +2.66++  1.Kf3 (323) 6
  5.00   0:00   +2.69   1.Kf3 Ke6 (351) 5
  6.00   0:00   +2.79   1.Kf3 Ke6 2.Kg3 (506) 8
  7.00   0:00   +2.99++  1.Kf3 (964) 12
  7.00   0:00   +3.19++  1.Kf3 (1.119) 12
  7.00   0:00   +3.19   1.Kf3 Ke6 2.Kg3 Kd5 3.Kh4 (1.181) 12
  8.00   0:00   +3.39++  1.Kf3 (2.115) 15
  8.00   0:00   +3.39   1.Kf3 Ke6 2.Kg3 Kd5 3.Kh4 Ke4 4.Kxh5 (2.255) 14
  9.00   0:00   +3.39   1.Kf3 Ke6 2.Kg3 Kd6 3.Kh4 (3.297) 16
10.00   0:00   +3.59++  1.Kf3 (4.834) 19
10.00   0:00   +3.69   1.Kf3 Ke6 2.Kg3 f5 3.Kh4 (6.712) 23
11.00   0:00   +3.89++  1.Kf3 (10.483) 26
11.00   0:00   +4.09++  1.Kf3 (12.789) 27
11.00   0:00   +4.49++  1.Kf3 (14.953) 28
11.00   0:00   +5.00   1.Kf3 Ke6 2.Kg3 f5 3.Kh4 Kd5 4.Kxh5 (19.214) 31
12.00   0:01   +5.11   1.Kf3 (58.581) 48
13.00   0:03   +5.12   1.Kf3 (229.096) 60
14.01   0:13   +5.11   1.Kf3 (893.483) 66
15.01   0:58   +5.12   1.Kf3 (4.105.417) 72

Best,
Orlando
Parent - By Ernest Bonnem Date 2010-03-18 22:19
[quote="Orlando de la Vega"]your results are very strange because you obtained two different analysis.[/quote]
Of course, this is MP (2-cpu) = not reproducible !!!
Parent - By Ernest Bonnem Date 2010-03-18 22:43
And here is my result with 1-cpu, in the Shredder GUI:
(in the Fritz GUI with 1-cpu, the result is different, because of the "Sampled Search")

Engine: Rybka 3 32-bit (512 MB)
by Vasik Rajlich, Larry Kaufman
13.00   0:01   +5.12   1.Kf3 (229.096) 124
14.01   0:06   +5.11   1.Kf3 (893.483) 131
15.01   0:29   +5.12   1.Kf3 (4.105.417) 141
16.01   0:44   +3.51   1.Kf3 Kc6 2.Kg3 Kb5 3.Kf4 Ka4 4.Ke5 Ka5 5.Ke4 (6.172.091) 142
17.01   0:58   +0.37   1.Kf3 Kc6 2.Kg3 Kb5 3.Kh4 Ka4 4.Kxh5 Kb3 (7.912.838) 138  TB:139
18.01   1:04   +0.37   1.Kf3 Kc6 2.Kg3 Kb5 3.Kh4 Ka4 4.Kxh5 Kb3 (8.514.307) 135  TB:214
19.01   1:09   +0.37   1.Kf3 Kc6 2.Kg3 Kb5 3.Kh4 Ka4 4.Kxh5 Kb3 (9.097.581) 133  TB:310
20.01   1:14   +0.37   1.Kf3 Kc6 2.Kg3 Kb5 3.Kh4 Ka4 4.Kxh5 Kb3 5.Kg4 Kxb2 6.a4 Kxc3 7.a5 Kb2 8.a6 c3 9.a7 c2 10.a8Q c1Q 11.Qb7+ Ka3 12.Qe7+ Kb2 13.Qf6+ Ka3 14.Qd6+ (9.550.353) 131  TB:442
21.01   1:26   +0.37   1.Kf3 Kc6 2.Kg3 Kb5 3.Kh4 Ka4 4.Kxh5 Kb3 5.Kg4 Kxb2 6.a4 Kxc3 7.a5 Kb2 8.a6 c3 9.a7 c2 10.a8Q c1Q 11.Qb7+ Ka3 12.Qe7+ Kb2 13.Qf6+ Ka3 14.Qd6+ (10.698.303) 127  TB:752
best move: Ke2-f3 time: 1:40.547 min  n/s: 123.431  nodes: 12.095.488 TB: 1.248

By the way, your system seems a little slow... 
Parent - By Benno Hartwig Date 2010-03-18 11:30
40/1'  und nur 50 Partien?
Hui, must du es aber eilig gehabt haben.
Benno
Up Topic Hauptforen / CSS-Forum / FireBird 1.2 against Rybka 3

Powered by mwForum 2.29.3 © 1999-2014 Markus Wichitill