Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / 'Famous' engines vs. modern top engines
- - By Kappatoo (*****) [us] Date 2017-09-22 14:11
If I'm not mistaken, the engine Kramnik lost to in 2006 was Deep Fritz 10. In the CCRL rating list, DF 10 has a rating of 2832, compared to Komodo 11.2's 3401. If I'm not mistaken, this gives DF 10 an expectancy of 0.036 against Komodo. (Is this correct?) Since these results are on equal hardware, I was wondering what this would translate to if DF 10 played on the hardware that was used in 2006.
Apart from DF 10, how do other engines that played matches against humans compare to today's top engines? The DF version that played against Kramnik in 2002, the version of Junior that played against Kasparov in 2003, etc. And are there reliable estimates about the strength of Deep Blue and Hydra?
Parent - - By Peter Grayson (****) [gb] Date 2017-09-22 21:09

> If I'm not mistaken, the engine Kramnik lost to in 2006 was Deep Fritz 10.


The engine only used 2 threads and was subsequently quietly, almost secretly updated. See ...
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=24574
Parent - By Kappatoo (*****) [us] Date 2017-09-22 21:44
Thanks! I vaguely remembered something like this.
Parent - By rocket (***) [se] Date 2017-09-29 12:49
It actually only used 1 thread. The bugg apparently made it run on single processor when it was set to 4.....
Parent - - By leavenfish (***) [us] Date 2017-09-23 16:05
Given that older engines were built in an age when hardware was limited - they were built for that hardware (!) it seems unfair to play one against another in a game. That said, I think testing individual positions might be fair - evaluation function vs evaluation function. But you still might need to limit the newer engine. How about this idea: let each engine in a game or position test, go to 17 ply and see who wins?
Parent - - By brunjes (**) [us] Date 2017-09-23 16:27
I am not sure that would be meaningful, since one ply of search depth by engine X is not necessarily the same as that same number of ply for engine Y.
Parent - By leavenfish (***) [us] Date 2017-09-23 23:27
I believe you were replying to me not the OP.
I'm not sure it would be meaningful either...but, how best to test? I can't see a good way that does not take into account the hardware an engine is running on and specifically now an engine might be designed to take advantage of it. I mean, even now I see how Komodo is supposedly performing better on the new AMD 'Zen' processessors vs some similar Intel processors.

I'm just not sure you can compare different generations of engines against each other...kind of like how you cannot truly compare Alekhine vs Karpov or Carlsen.
Parent - By Kappatoo (*****) [us] Date 2017-09-23 16:32
I didn't mean to suggest it was fair. I am just trying to gauge how strong Kasparov's and Kramnik's (etc.) opponents were in comparison with today's engines.
- - By rocket (***) [se] Date 2017-09-29 12:53 Edited 2017-09-29 12:59
Just because the threadmaker asked...

I will play tournament time controls (120/40) 1 CPU between Deep Fritz 10.1 and Stockfish 8.

Let's say 20 game match.:cool:

I predict 19 wins, 1 draw for S8
Parent - - By Kappatoo (*****) [us] Date 2017-09-29 18:49
Cool! Keep us posted.
Parent - By rocket (***) [se] Date 2017-10-02 10:38
Game 1

[Event "DESKTOP-3NMBD1O, 120'/40"]
[Site "DESKTOP-3NMBD1O"]
[Date "2017.09.29"]
[Round "1"]
[White "Deep Fritz 10"]
[Black "Stockfish 8 64 POPCNT"]
[Result "0-1"]
[ECO "C97"]
[Annotator "0.11;0.37"]
[PlyCount "93"]
[TimeControl "40/7200:0/0:0/0"]

{Intel(R) Celeron(R) CPU  N3050  @ 1.60GHz  W=18.5 plies; 1 066kN/s; 67 TBAs
B=33.6 plies; 625kN/s} 1. e4 {Black last book move} e5 2. Nf3 Nc6 3. Bb5 a6 4.
Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 d6 8. c3 O-O 9. h3 Na5 10. Bc2 c5 11. d4 {
White last book move} Qc7 {0.37/29 315} 12. Bd3 {0.11/15 250 (b3)} Bd7 {0.28/
27 247 (Be6)} 13. Be3 {0.30/16 430} h6 {0.15/27 120 (Nc4)} 14. Nbd2 {0.29/16
180} Rfe8 {0.31/26 68 (Rfc8)} 15. Rc1 {0.30/16 273} Rac8 {0.15/30 128 (Be6)}
16. b4 {0.39/16 179} cxb4 {0.11/33 267} 17. cxb4 {0.37/18 377} Qxc1 {0.08/32 49
} 18. Qxc1 {0.30/17 68} Rxc1 {0.10/36 307} 19. Rxc1 {0.29/18 107} Nc6 {0.21/34
35} 20. dxe5 {0.20/18 204} dxe5 {0.12/35 131 (Nxe5)} 21. a3 {0.28/19 287} Rc8 {
0.09/33 41 (Rd8)} 22. Nb3 {0.37/20 387} Nh5 {0.16/34 231 (Be6)} 23. Nc5 {0.52/
18 299} Bxc5 {0.06/32 162} 24. Rxc5 {0.53/18 105} Nf4 {0.10/31 65} 25. Bxf4 {
0.46/18 187 (Bb1)} exf4 {0.00/36 212} 26. Bb1 {0.40/19 478 (e5)} f6 {0.00/34
224} 27. Ba2+ {0.21/18 259 (Kf1)} Kf8 {0.00/38 225} 28. Bd5 {0.23/19 235 (Kf1)}
Ne7 {-0.14/37 263 (g5)} 29. Bb3 {0.23/19 397 (Rxc8+)} Rxc5 {-0.73/32 145} 30.
bxc5 {0.26/18 80} Nc6 {-0.56/32 72} 31. Bd5 {0.17/19 170 (Kf1)} a5 {-0.93/33
214 (Ke7)} 32. Kf1 {0.06/18 229} b4 {-0.60/32 77} 33. Nd2 {-0.20/19 371 (axb4)}
bxa3 {-1.46/34 207} 34. Nb1 {-0.29/20 244} a2 {-1.51/34 67 (f5)} 35. Bxa2 {-0.
09/19 294} Ne5 {-1.57/37 536} 36. Bd5 {-0.07/19 292 (Ke2)} Bb5+ {-1.83/34 243
(f3)} 37. Kg1 {-0.55/21 228} Ke7 {-1.78/32 77 (Ba6)} 38. Nc3 {-0.39/19 243} Ba6
{-2.02/35 428 (Bd3)} 39. Bb3 {-0.62/19 249 (Na4)} Nd3 {-2.20/38 424} 40. Nd5+ {
-0.63/18 56} Kd8 {-2.20/37 197} 41. c6 {-0.83/20 230} Bb5 {-2.20/37 348} 42.
c7+ {-1.11/19 190} Kc8 {-2.23/35 66} 43. f3 {-1.14/18 157 (Bd1)} Bc6 {-2.75/34
230} 44. Kf1 {-1.44/18 135 (Ne7+)} Bxd5 {-5.50/35 172} 45. Bxd5 {-1.68/22 296
(exd5)} a4 {-7.57/36 222 (Kxc7)} 46. Ke2 {-1.53/21 128} Nb4 {-8.78/37 176 (Ne5)
} 47. Be6+ {-1.83/22 179 (Bc4) adjud.} 0-1
- - By rocket (***) [se] Date 2017-09-29 13:00
Due to modern scaling improvements, only 1 CPU comparisons between engines of different generations are meaningful.
Parent - - By Vegan (****) [ca] Date 2017-10-03 18:14
I have been using multicore processors for over 10 years.....
Parent - By rocket (***) [se] Date 2017-10-04 08:42
The modern engines probably scale better...
- By rocket (***) [se] Date 2017-10-05 22:52
Tool lazy right now to conduct the rest of the match but this is a great game between the two:

[Event "DESKTOP-3NMBD1O, Blitz:5'+3""]
[Site "DESKTOP-3NMBD1O"]
[Date "2017.10.06"]
[Round "2"]
[White "Deep Fritz 10"]
[Black "Stockfish 8 64 POPCNT"]
[Result "0-1"]
[ECO "E97"]
[Annotator "0.49;0.45"]
[PlyCount "115"]
[TimeControl "300+3"]

{Intel(R) Celeron(R) CPU  N3050  @ 1.60GHz  W=14.7 plies; 1 349kN/s; 2 331
TBAs  B=23.3 plies; 731kN/s} 1. d4 {White last book move Black last book move}
Nf6 2. Nf3 g6 3. c4 Bg7 4. Nc3 O-O 5. e4 d6 6. Be2 Nbd7 7. O-O e5 8. dxe5 {0.
49/13 16} Nxe5 {0.45/19 17 (dxe5)} 9. Nxe5 {0.44/13 16} dxe5 {0.45/22 9} 10.
Be3 {0.39/13 15 (Qa4)} c6 {0.31/21 8} 11. Qa4 {0.43/13 12 (Bc5)} Be6 {0.30/19
10 (Re8)} 12. Rad1 {0.46/14 12 (a3)} Qe7 {0.12/19 5 (Qc7)} 13. Qa5 {0.55/14 15
(a3)} b6 {0.26/22 16} 14. Qxe5 {0.46/14 10} Nd5 {0.14/23 13 (Qb4)} 15. Nxd5 {
0.99/15 10} cxd5 {0.00/22 12} 16. Bg5 {0.60/16 18} Qc5 {-0.10/21 7 (Qb4)} 17.
Bf6 {0.87/16 15} Bxf6 {-0.06/21 2} 18. Qxf6 {0.87/15 5} dxc4 {0.00/22 11 (dxe4)
} 19. Rd4 {0.96/13 11 (Bf3)} Rac8 {-0.25/21 9 (Rae8)} 20. Rfd1 {0.85/14 11} c3
{-0.16/23 12 (Rfe8)} 21. bxc3 {0.91/14 12} Bxa2 {-0.31/23 20} 22. c4 {0.70/13
11 (Ba6)} Bb3 {-0.37/22 12} 23. R1d2 {0.61/14 11} a5 {-0.16/18 2 (Rfe8)} 24. h4
{0.38/13 16} Rce8 {-0.11/23 11} 25. Rd5 {0.52/13 8} Qe7 {0.00/22 3 (Qa3)} 26.
Qxb6 {0.57/15 10} a4 {-0.08/24 7} 27. e5 {0.72/14 7 (h5)} Qxh4 {-0.29/20 7} 28.
Qe3 {0.54/14 8} Qe7 {-0.26/20 2} 29. f4 {0.62/14 15} Rd8 {0.00/23 19 (Qb4)} 30.
R2d3 {0.67/12 10 (f5)} Ra8 {0.00/21 30} 31. Rd7 {0.39/12 6} Qb4 {0.00/26 11}
32. Rc3 {0.31/13 5 (Bf3)} a3 {-0.82/22 6} 33. Rxb3 {0.06/15 9} Qe1+ {-0.76/20 2
} 34. Bf1 {-0.05/17 5} Qxe3+ {-0.72/24 25} 35. Rxe3 {-0.05/17 2} a2 {-0.67/21 3
} 36. Rd1 {-0.09/17 13 (Re1)} a1=Q {-0.93/23 5} 37. Rxa1 {-0.09/15 3} Rxa1 {-0.
93/22 5} 38. Kf2 {-0.12/15 8} f6 {-0.98/23 11 (Rd8)} 39. e6 {0.02/14 7 (exf6)}
Rb8 {-0.93/21 11 (Rc8)} 40. Be2 {0.01/14 8} Ra2 {-0.87/23 10 (f5)} 41. Kg3 {0.
02/14 6} Ra7 {-0.92/23 8} 42. Kh4 {0.00/14 6 (f5)} Kf8 {-1.25/21 4} 43. Bf3 {
0.00/13 4} Ke7 {-1.16/17 1} 44. g4 {-0.22/13 6 (Bd5)} Rb4 {-1.40/22 5 (Rb1)}
45. Bd5 {-0.19/15 7} Rba4 {-1.65/23 4 (Rb1)} 46. Rb3 {-0.37/14 10 (f5)} Ra3 {
-1.70/28 11 (Ra2)} 47. Rg3 {-0.58/13 5} Rxg3 {-1.87/26 2 (g5+)} 48. Kxg3 {-0.
64/14 1} Ra3+ {-1.80/32 11} 49. Kf2 {-1.04/17 6} Rc3 {-1.80/36 7} 50. Kg2 {-1.
32/17 9} h6 {-1.80/36 2} 51. Kf2 {-1.10/15 4 (Be4)} f5 {-3.65/26 4} 52. gxf5 {
-1.68/17 5} gxf5 {-3.94/24 2} 53. Kg2 {-1.68/18 5 (Ke2)} h5 {-4.53/25 3} 54.
Bf3 {-1.65/18 3 (Kh2)} h4 {-5.91/29 10} 55. Bd5 {-2.19/18 5} Rg3+ {-6.26/26 4
(Rd3)} 56. Kh2 {-2.61/18 3} Kd6 {-6.91/27 4} 57. Kh1 {-2.66/18 5} Rg4 {-7.39/
27 4} 58. Kh2 {-3.19/17 4 adjud.} 0-1
Up Topic The Rybka Lounge / Computer Chess / 'Famous' engines vs. modern top engines

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill