Uniform Platform Match "DarkThought" vs. "XXXX II"
Ernst A. Heinz
************************************************************
Report on Uniform Platform Match "DarkThought" vs. "XXXX II"
************************************************************
Hi all,
From January 9-11, 1998, the two amateur chess programs "DarkThought" and
"XXXX II" -- both participants of the last three ICCA World Microcomputer
Chess Championships -- played a uniform platform match at the Institute for
Program Structures and Data Organization, University of Karlsruhe, Germany.
Two 500MHz DEC Alpha-21164a PC164 workstations with 256MB RAM and Digital
Unix 4.0b system software as provided by Digital Equipment Corporation under
external technology research contract DE-0032-97 served as the uniform
platform hosts.
Initially, the author of "XXXX II" (Martin Zentner) proved that his C++
program is very well designed. Helped by the authors of "DarkThought"
(Markus Gille, Ernst A. Heinz) he ported "XXXX II" to the 64bit DEC Unix
system in a couple of hours. Although this was the first time for "XXXX II"
to use a true 64bit environment it performed flawlessly during the whole
match. Using the DEC Unix C++ compiler with heavy optimizations enabled,
"XXXX II" ran roughly 10% faster on the 500MHz DEC Alpha-21164a than on a
300MHz Pentium-II under Windows/NT when compiled with Microsoft Visual C++.
The match itself consisted of 9 games at tournament time controls (30 moves
in 60 minutes, then repeated 40 moves in 60 minutes). The first five games
were played from the so-called ``Nunn Positions #4 and #5'' with opening
books disabled. As intended by IGM John Nunn, both programs felt about
equal in the starting positions. "DarkThought" won all these five games
although it accidentally played Black with a tempo less than required in
the very first game. With enabled opening books the games would have
progressed completely different because both programs found the starting
positions in their books. Furthermore, the Nunn game setup seems to favor
the faster program (in this case "DarkThought") as the first 10-12 common
theory moves do not contribute to the time control.
The last four games were normal tournament games with opening books
enabled. Here, it quickly became obvious that "XXXX II" features a much
better opening library than "DarkThought" which had to fight really hard
in order to win three and lose only one of the normal games.
The overall score of the match: "DarkThought" 8 points,
"XXXX II" 1 point.
Please find the notations of all games below.
=Ernst A. Heinz= & =Martin Zentner=
P.S. Visit the WWW page of "DarkThought" at
URL = <http://wwwipd.ira.uka.de/Tichy/DarkThought>.
**********
Nunn #4 (accidentally with wrong side to move)
FEN = r1b2rk1/2q1bppp/p2p1n2/npp1p3/3PP3/2P2N1P/PPBN1PP1/R1BQR1K1 w - - 0 1
XXXX II - DarkThought (time = 30 60 40 60)
12.. % 13. dxe5 dxe5 14. b3 Rd8 15. Qe2 Nh5 16. Qe3 Nf4 17. Ba3 Bxh3
18. gxh3 Qc8 19. Nxe5 Bg5 20. Kh1 Rxd2 21. h4 Bh6 22. Nxf7 Kxf7 23. Bb1 Kg8
24. Qg3 Nh3 25. Rf1 Bf4 26. Qf3 Nxf2+ 27. Rxf2 Rxf2 28. Qxf2 Qh3+ 29. Kg1 Rf8
30. Bd3 Be3 31. Qxe3 Qxe3+ 32. Kh1 Rf2 33. Bc4+ bxc4 34. Rd1 Qxe4+
35. Kg1 Qg2# 0-1
**********
Nunn #4 (Black to move)
1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O b5 6. Bb3 Be7 7. Re1 O-O
8. c3 d6 9. h3 Na5 10. Bc2 c5 11. d4 Qc7 12. Nbd2
FEN = r1b2rk1/2q1bppp/p2p1n2/npp1p3/3PP3/2P2N1P/PPBN1PP1/R1BQR1K1 b - - 0 1
DarkThought - XXXX II (time = 30 60 40 60)
12.. exd4 13. cxd4 Nc6 14. b3 cxd4 15. Bb2 Nb4 16. Nxd4 Bd7 17. Bb1 Rfe8
18. Nf1 Rad8 19. Ne3 Nc6 20. Ndf5 Be6 21. Qc1 Qa5 22. Rd1 Ne5 23. Nxe7+ Rxe7
24. Qd2 Qxd2 25. Rxd2 Ree8 26. f4 Nc6 27. f5 Bc8 28. Bxf6 gxf6 29. Nd5 Kg7
30. Rd3 Re5 31. a4 Bb7 32. axb5 axb5 33. Kh2 Kh8 34. Nxf6 Rc5 35. Rd5 Ne5
36. Rd1 Bc6 37. Ra7 Rc3 38. Ng4 Nxg4+ 39. hxg4 Be8 40. e5 d5 41. Be4 Rxb3
42. e6 fxe6 43. fxe6 Bg6 44. Bxg6 hxg6 45. Rc1 Re8 46. Rcc7 Re3 47. Rh7+ Kg8
48. Rag7+ Kf8 49. Rf7+ Kg8 50. e7 R3xe7 51. Rxe7 Rxe7 52. Rxe7 1-0
XXXX II - DarkThought (time = 30 60 40 60)
12.. cxd4 13. cxd4 Bb7 14. d5 Rac8 15. Bd3 Nd7 16. Nf1 f5 17. exf5 Bxd5
18. Ng5 Bxg5 19. Bxg5 Nb6 20. Ne3 Nac4 21. Nxd5 Nxd5 22. Qe2 Kh8 23. Rac1 Qf7
24. Bxc4 bxc4 25. Rxc4 Nf4 26. Rxf4 exf4 27. Qxa6 Qxf5 28. Be7 Rfe8 29. Qxd6 f3
30. gxf3 Qxh3 31. b4 Qxf3 32. Qd7 Qf7 33. a3 Rc2 34. Qa7 Qg6+ 35. Kh2 Qf5
36. Rf1 Qf4+ 37. Kg2 Rc6 38. Qe3 Rg6+ 39. Qg3 Qe4+ 40. f3 Rxg3+ 41. Kxg3 Qxe7
42. f4 Qe3+ 43. Rf3 Qg1+ 44. Kh4 Re2 45. Rg3 Qh2+ 46. Kg4 h5+ 47. Kf5 Qxg3
48. a4 Qg4# 0-1
**********
Nunn #5 (White to move)
1. d4 d5 2. c4 c6 3. Nf3 e6 4. cxd5 exd5 5. Nc3 Nf6 6. Bg5 Be7 7. Qc2 Nbd7
8. e3 O-O 9. Bd3 Re8 10. O-O Nf8
FEN = r1bqrnk1/pp2bppp/2p2n2/3p2B1/3P4/2NBPN2/PPQ2PPP/R4RK1 w - - 0 1
DarkThought - XXXX II (time = 30 60 40 60)
11. Rab1 Bd6 12. b4 h6 13. Bh4 a6 14. Na4 Bg4 15. Nc5 Bxf3 16. gxf3 Rb8
17. a4 Be7 18. Kh1 Nh5 19. Bxe7 Rxe7 20. b5 axb5 21. axb5 Nf6 22. bxc6 bxc6
23. Rxb8 Qxb8 24. Rc1 Qe8 25. Na6 Ne6 26. Qxc6 Qxc6 27. Rxc6 Rb7 28. Bf5 Nf8
29. Kg2 Ra7 30. Nc5 Ra8 31. Rd6 Rb8 32. Bc2 Ng6 33. Nd7 Nxd7 34. Rxd7 Rb5
35. Ba4 Ra5 36. Bc6 Nf8 37. Rxd5 Rxd5 38. Bxd5 Nd7 39. f4 Nb6 40. Bb3 Nc8
41. e4 Nd6 42. Kf3 Nb5 43. Ke3 Nd6 44. e5 Nb7 45. f5 Kf8 46. d5 Na5 47. Ba4 f6
48. f4 Nc4+ 49. Kd4 Nb6 50. Bb5 Ke7 51. Kc5 1-0
XXXX II - DarkThought (time = 30 60 40 60)
11. Rac1 Ng6 12. Rfe1 Be6 13. Ne5 Bd6 14. f4 Bc8 15. Nf3 Bb4 16. f5 Nf8
17. Qb3 Be7 18. e4 dxe4 19. Nxe4 Nd5 20. Bd2 Nd7 21. Re2 N7b6 22. Nc3 a5
23. Na4 Nxa4 24. Qxa4 Bd7 25. Rce1 Bb4 26. Rxe8+ Bxe8 27. Bxb4 axb4 28. Qb3 f6
29. Nd2 Bf7 30. Bb1 Qb6 31. Qd3 Nc7 32. Nb3 Qa7 33. Re7 Nd5 34. Re1 b6
35. Nd2 Qd7 36. Nb3 Qd6 37. Nd2 Nf4 38. Qe3 Rd8 39. Nf3 Kf8 40. g3 Nd5
41. Qe4 Nc7 42. Qg4 Nb5 43. Qh4 Nxd4 44. Nxd4 Qxd4+ 45. Qxd4 Rxd4 46. Kf2 c5
47. Ke3 Rd8 48. Rc1 Rd5 49. h3 Rd4 50. h4 Rd7 51. b3 Bh5 52. Be4 Bg4 53. Rc2 h5
54. Rd2 Ke7 55. Rxd7+ Kxd7 56. Kf4 Kd6 57. Bg2 Be2 58. Ke3 Bd1 59. Kf4 Bc2
60. Bf3 Bb1 61. Bxh5 Bxa2 62. Bf7 b5 63. Ke3 c4 64. bxc4 bxc4 0-1
**********
Normal Tournament Games
DarkThought - XXXX II (time = 30 60 40 60)
1. e4 e5 2. f4 Bc5 3. Nf3 d6 4. Nc3 Nc6 5. Bb5 Bg4 6. Na4 exf4 7. d4 Bxf3
8. gxf3 Bb4+ 9. c3 Ba5 10. b4 Qh4+ 11. Ke2 Bb6 12. Nxb6 axb6 13. d5 Nge7
14. Kf1 f5 15. Qe2 Kf8 16. dxc6 Nxc6 17. Rg1 Re8 18. Qd2 g5 19. exf5 Re5
20. Bd3 Rd5 21. Qe2 Qh3+ 22. Rg2 Re5 23. Qf2 Ne7 24. c4 Nxf5 25. Bxf5 Rxf5
26. Bb2 g4 27. Qd4 Rg8 28. Kg1 g3 29. Kh1 Rg6 30. Re1 Rff6 31. Bc3 Qc8
32. Rge2 Qh3 33. Re8+ Kg7 34. Qd2 c6 35. Qb2 Kh6 36. Bxf6 1-0
DarkThought - XXXX II (time = 30 60 40 60)
1. e4 e5 2. f4 exf4 3. Nf3 g5 4. h4 g4 5. Ne5 Nf6 6. d4 d6 7. Nd3 Nxe4
8. Qe2 f5 9. Nc3 Qe7 10. Nxe4 Qxe4 11. Nxf4 Nc6 12. Qxe4+ fxe4 13. Be3 Ne7
14. Nh5 Nf5 15. Bg5 Be7 16. O-O-O Kd8 17. Be2 Bxg5+ 18. hxg5 Ne3 19. Rdg1 Be6
20. Nf6 Bxa2 21. b3 Nf5 22. Bxg4 Nxd4 23. g6 Bxb3 24. cxb3 Ke7 25. Nd5+ Kf8
26. Rf1+ Kg7 27. Rf7+ Kxg6 28. Bh5+ Kh6 29. Re7 Nxb3+ 30. Kb2 Kg5 31. Rg7+ Kf5
32. Bg4+ Ke5 33. Rd1 Nd4 34. Rg5+ Nf5 35. Rxf5+ Ke6 36. Rf6+ Ke5 37. Re6# 1-0
XXXX II - DarkThought (time = 30 60 40 60)
1. e4 a6 2. d4 d5 3. exd5 Qxd5 4. Nf3 Bg4 5. Nc3 Qd6 6. h3 Bh5 7. g4 Bg6
8. h4 h5 9. Ne5 Nc6 10. Nxg6 Qxg6 11. gxh5 Qd6 12. d5 Ne5 13. Bg5 Nf6
14. Bg2 Qb4 15. h6 gxh6 16. Bxf6 exf6 17. Rb1 Bc5 18. Qe2 O-O-O 19. Kf1 Qf4
20. Ne4 Bb6 21. Qd2 Qxd2 22. Nxd2 Ng4 23. Bf3 f5 24. Nc4 Ba7 25. Bxg4 fxg4
26. Ne5 Rxd5 27. Nxg4 f5 28. Ne3 Rd2 29. Rc1 Rhd8 30. Rh3 f4 31. Rf3 fxe3
32. fxe3 Rh2 33. Kg1 Rdd2 34. b3 Rxh4 35. c4 Rxa2 36. c5 Kd7 37. Rc3 Kc6
38. Rf6+ Kb5 0-1
XXXX II - DarkThought (time = 30 60 40 60)
1. e4 e6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Bd7 5. Nf3 Bc6 6. Bd3 Nd7 7. c3 Be7
8. O-O Ngf6 9. Nxf6+ Bxf6 10. Bf4 O-O 11. Qe2 Bd5 12. Rad1 Kh8 13. c4 Bxf3
14. Qxf3 Qc8 15. Be4 c6 16. Qh3 g6 17. Qf3 Re8 18. Bg3 Qd8 19. d5 exd5
20. cxd5 c5 21. d6 Bxb2 22. Bxb7 Rb8 23. Bc6 f6 24. Qd5 Qc8 25. Bxd7 Qxd7
26. Qxc5 Rbc8 27. Qd5 Bc3 28. Rc1 Rf8 29. Rfd1 g5 30. Rc2 a6 31. Qd3 Ba5
32. Rxc8 Rxc8 33. Qxa6 Bd8 34. Qd3 Qa4 35. Ra1 Rc4 36. a3 Qc6 37. Qd2 Kg7
38. Re1 h5 39. h4 gxh4 40. Bf4 Ra4 41. Qe3 Qb5 42. Bh6+ Kh7 43. Bf8 Rg4
44. Qf3 Rg8 45. Be7 Qg5 46. Re6 h3 47. g3 h4 48. Kh2 hxg3+ 49. fxg3 Bxe7
50. Rxe7+ Kg6 51. Qe4+ Qf5 52. Qxf5+ Kxf5 53. d7 Rd8 54. Kxh3 Kg5 55. g4 f5
56. Re5 Rxd7 57. Rxf5+ Kg6 58. a4 Rd1 59. a5 Rh1+ 60. Kg3 Rg1+ 61. Kf3 Rf1+
62. Ke4 Rg1 63. Rb5 Rxg4+ 64. Kd5 Kh6 65. a6 Rg5+ 66. Kc4 Rg7 67. Rc5 Rg1
68. a7 Ra1 69. Rc7 Rc1+ 70. Kd5 Rxc7 71. a8=Q Rf7 72. Ke6 Rf4 73. Ke5 Rg4
74. Qh8+ Kg5 75. Qh3 Rb4 76. Qe3+ Kg4 77. Qe2+ Kh3 78. Kf5 Rb3 79. Kf4 Rb8
80. Qd3+ Kg2 81. Qa3 Rd8 82. Qb2+ Kh1 83. Qb7+ Kh2 84. Qc7 Rf8+ 85. Ke4+ Kh1
86. Qh7+ Kg1 87. Qg7+ Kh1 88. Qxf8 Kg2 89. Qf3+ Kg1 1-0
**********
Komputer Korner
Is the difference really only 10% in speed? If that is so, then for chess
programs, DEC Alphas are not cost effective
--
- -
Komputer Korner
The inkompetent komputer
If you see a 1 in my email address, take it out before replying.
Please do not email both me and the r.g.c.c. at the same time. I read all
the postings on r.g.c.c.
Also every statement of mine should be taken with a grain of salt. Read at
your own risk and
assume that it is only this humble komputer's opinion.
Ernst A. Heinz <hei...@ira.uka.de> wrote in article
<69l282$bb3$1...@nz12.rz.uni-karlsruhe.de>...
bruce moreland
On 15 Jan 1998 19:59:43 EST, "Komputer Korner" <kor...@netcom.ca>
wrote:
>Is the difference really only 10% in speed? If that is so, then for chess
>programs, DEC Alphas are not cost effective
Probably a little more in the typical case.
Also, some of them are faster than others, it depends upon more stuff
than just the clock speed.
I know of a couple of chess programs that get 25-30% more on a 533,
compared to a P2/300, without being specially engineered to use 64
bits.
Buying an alpha is a little strange to start with, since it will not
run Windows '95.
bruce
Ernst A. Heinz
>> Is the difference really only 10% in speed? If that is so, then for chess
>> programs, DEC Alphas are not cost effective
>
> Probably a little more in the typical case.
>
> Also, some of them are faster than others, it depends upon more stuff
> than just the clock speed.
Absolutely correct.
Our 500MHz DEC Alpha-21164a PC164 systems contain the "old" PC164 mainboards
with just 1MB L3 cache and standard 70ns PS/2 SIMMs (fast page mode).
The new LX164 mainboards are much better! They feature a new chipset, at least
2MB L3 cache and SDRAM-DIMMs such that memory-intensive programs will probably
get another 10% from using the LX164 mainboards.
> I know of a couple of chess programs that get 25-30% more on a 533,
> compared to a P2/300, without being specially engineered to use 64
> bits.
Yes, this agrees with our experience of the new LX164 mainboards.
=Ernst=