Myrddin v0.88 Release

[JVMerlino]

Hello All,

After almost exactly 6.5 years, there's a new release of Myrddin, version 0.88. Testing estimates an increase of about 110-115 elo at 1 CPU, plus slight efficiency improvement with my Very Lazy SMP (previous versions were about +90 at 4 CPU and this is around +100).

Many thanks to all who kept me interested over the long break, particularly Martin Sedlak (Cheng) who guided me through tuning, Graham Banks who regularly expressed a desire for a new version, and Lars Hallerstrom for his invaluable testing.

You can download the package here:
http://computer-chess.org/doku.php?id=c ... ddin:index

jm

[Rebel]

Hi John, good to see you active again.

I put Myrddin 0.88 on the rack for the Gambit Rating List.

1400 games.

You can follow the progress here - http://rebel13.nl/a/grl.htm

Page automatically refreshes each 30 seconds.

Good start after 19 games.

[JVMerlino]

Hi John, good to see you active again.

I put Myrddin 0.88 on the rack for the Gambit Rating List.

1400 games.

You can follow the progress here - http://rebel13.nl/a/grl.htm

Page automatically refreshes each 30 seconds.

Good start after 19 games.

Thanks very much, Ed!

[Gabor Szots]

This must be an April's Fool joke. But if true, which I seriously doubt, thanks John.

[JVMerlino]

This must be an April's Fool joke. But if true, which I seriously doubt, thanks John.

I know. I can't believe it either!

[Rebel]

Code: Select all

Gambit Rating List
Running      : gauntlet Myrddin 0.88
Time Control : Time control 40/120
Games        : 1400

Results from file gauntlet-Myrddin.pgn:

No. Name           Win Draw Loss Unf.  Score Games       %
----------------------------------------------------------
  1 Myrddin 0.88  +699 =348 -353   *0  873.0  1400   62.4%
  2 Zahak 4.0      +50  =77  -73   *0   88.5   200   44.2%
  3 Jumbo 0.6.10   +60  =49  -91   *0   84.5   200   42.2%
  4 Monolith 0.3   +53  =51  -96   *0   78.5   200   39.2%
  5 Orion 04       +48  =55  -97   *0   75.5   200   37.8%
  6 Foxsee 7.20.1  +53  =44 -103   *0   75.0   200   37.5%
  7 CT800 1.42     +49  =36 -115   *0   67.0   200   33.5%
  8 Jumbo 0.5.3    +40  =36 -124   *0   58.0   200   29.0%

Total Games:    1400
White Wins:      513 (36.6%)
Black Wins:      539 (38.5%)
Draws:           348 (24.9%)
Unfinished:        0 (0.0%)

Estimated ratings for this elo 2489 pool

   # PLAYER           :  RATING  POINTS  PLAYED   (%)
   1 Myrddin 0.88     :  2567.3   873.0    1400    62
   2 Zahak 4.0        :  2526.8    88.5     200    44
   3 Jumbo 0.6.10     :  2512.5    84.5     200    42
   4 Monolith 0.3     :  2490.7    78.5     200    39
   5 Orion 04         :  2479.6    75.5     200    38
   6 Foxsee 7.20.1    :  2477.7    75.0     200    38
   7 CT800 1.42       :  2447.1    67.0     200    34
   8 Jumbo 0.5.3      :  2410.3    58.0     200    29

Estimated elo 2567

[JVMerlino]

Estimated elo 2567

Excellent. Thanks, again!

[Gabor Szots]

It seems there is a problem with repetition detection. Here is a sample game (played under Arena 3.5.1):

[pgn][Event "Myrddin 0.88 64-bit - Jul 19"]
[Site "Szentendre"]
[Date "2021.07.19"]
[Round "6"]
[White "Yace 0.99.87"]
[Black "Myrddin 0.88 64-bit"]
[Result "1-0"]
[ECO "D32"]
[Opening "QGD Tarrasch"]
[Time "19:46:20"]
[Variation "von Hennig-Schara Gambit, 7.Qxd5 Nc6 8.Bg5"]
[TimeControl "120+1"]
[PlyCount "171"]

1. d4 d5 2. c4 e6 3. Nc3 c5 4. cxd5 cxd4 5. Qa4+ {+0.74/10 8} Bd7 {-0.25/15
4} 6. Qxd4 {+0.65/11 7} exd5 {-0.39/14 3} 7. Qxd5 {+0.65/10 6} Nc6
{-0.34/13 3} 8. Bg5 {+0.50/10 5} Nf6 {-0.33/13 3} 9. Qd2 {+0.55/9 2} Be7
{-0.34/12 3} 10. Bxf6 {+0.50/9 2} Bxf6 {+0.13/13 3} 11. Rd1 {+0.28/9 4} Be6
{+0.07/13 3} 12. Qxd8+ {+0.10/9 3} Rxd8 {+0.49/15 3} 13. Rxd8+ {+0.07/10 3}
Kxd8 {+0.44/16 3} 14. a3 {+0.18/11 3} Kc7 {+0.44/14 3} 15. e4 {+0.36/9 1}
Re8 {+0.42/13 5} 16. Nge2 {+0.31/10 3} Bb3 {+0.32/12 3} 17. f4 {+0.53/11 3}
Rd8 {+0.41/12 3} 18. e5 {+0.48/11 3} Be7 {+0.36/13 3} 19. Nc1 {+0.49/10 3}
Be6 {+0.38/13 3} 20. Bd3 {+0.39/10 3} Nd4 {+0.29/12 4} 21. O-O {+0.67/10 3}
Nf5 {+0.35/12 2} 22. b4 {+0.81/10 2} h5 {+0.16/11 2} 23. Be4 {+0.87/10 2}
Kb8 {+0.13/12 2} 24. Re1 {+0.81/9 1} Rd4 {+0.03/12 2} 25. Nd3 {+0.82/9 2}
Rc4 {+0.12/14 2} 26. Na4 {+0.67/10 2} b6 {+0.35/14 3} 27. g3 {+0.64/10 2}
g6 {+0.50/13 2} 28. Ndb2 {+0.66/9 1} Rc8 {+0.48/14 2} 29. Nd1 {+0.43/10 2}
h4 {+0.56/14 2} 30. g4 {+0.68/10 1} Nd4 {+0.23/14 2} 31. Ne3 {+0.36/10 2}
Bd7 {+0.20/13 2} 32. Nb2 {+0.85/9 2} h3 {+0.08/13 2} 33. Rd1 {+0.89/10 2}
Ne2+ {+0.08/14 2} 34. Kf1 {+0.83/11 2} Bb5 {-0.04/15 2} 35. Bd3 {+0.92/11
2} Bxd3 {-0.39/16 6} 36. Nxd3 {+0.93/11 2} Nc3 {-0.39/16 2} 37. Rc1
{+0.92/11 2} a5 {-0.35/15 3} 38. g5 {+0.94/10 2} axb4 {-0.22/14 2} 39. Nxb4
{+0.88/10 1} Kb7 {-0.14/15 2} 40. Rc2 {+0.85/10 1} Ne4 {-0.45/15 4} 41. Rb2
{+0.83/10 2} Bc5 {+0.04/13 1} 42. Ned5 {+0.74/10 2} Bg1 {0.00/13 3} 43. Nd3
{+0.90/10 2} Rd8 {-0.02/13 1} 44. Kxg1 {+1.09/11 1} Rxd5 {-0.16/16 1} 45.
Nf2 {+0.98/11 1} Nxf2 {-0.26/17 1} 46. Kxf2 {+1.00/10 2} Rd3 {-0.25/17 2}
47. Re2 {+1.38/12 1} Rxa3 {-0.45/17 2} 48. e6 {+1.18/12 1} fxe6 {-0.44/17
1} 49. Rxe6 {+1.33/13 1} b5 {-0.34/16 2} 50. Rxg6 {+1.20/12 1} b4 {-0.34/16
2} 51. Re6 {+1.22/12 1} b3 {-0.21/15 2} 52. Re1 {+1.19/12 2} Kc7 {-0.30/15
1} 53. Rb1 {+1.13/12 2} Kd6 {-0.32/16 1} 54. Kf3 {+1.18/13 2} b2+ {-0.77/17
1} 55. Kg4 {+1.09/15 1} Rb3 {-0.85/19 1} 56. g6 {+1.28/15 1} Ke7 {-0.78/19
1} 57. Kg5 {+1.28/15 1} Rb4 {-0.32/20 7} 58. f5 {+1.25/14 1} Rb5 {-0.32/20
1} 59. g7 {+0.97/14 1} Kf7 {-0.32/20 1} 60. Kh6 {+0.92/14 1} Rb7 {-0.32/18
1} 61. f6 {+0.92/14 1} Rb4 {0.00/18 2} 62. Rg1 {+0.89/13 1} Rh4+ {0.00/17
1} 63. Kg5 Rb4 {0.00/17 1} 64. Kh5 {+0.89/13 1} Rb8 {0.00/16 1} 65. Rb1
{+0.89/14 1} Rb4 {0.00/19 1} 66. Kh6 {+0.83/13 1} Rh4+ {0.00/19 1} 67. Kg5
Rb4 {0.00/19 1} 68. Kf5 {+0.89/13 1} Rb8 {0.00/19 1} 69. Kf4 {+0.89/14 1}
Kxf6 {0.00/20 1} 70. Rxb2 {+0.98/14 1} Rg8 {0.00/23 1} 71. Rb3 {+0.98/14 0}
Rxg7 {0.00/27 1} 72. Rxh3 {+0.95/14 1} Rg8 {0.00/129 2} 73. Rh6+ {+1.01/12
1} Kf7 {0.00/129 2} 74. h4 {+1.11/13 1} Kf8 {0.00/129 2} 75. h5 {+1.16/13
1} Ke7 {0.00/129 2} 76. Ra6 {+1.25/13 1} Rh8 {0.00/129 2} 77. h6 {+1.43/14
1} Kf7 {0.00/129 2} 78. Kf5 {+1.40/15 1} Rd8 {0.00/129 2} 79. Rf6+
{+1.37/15 1} Kg8 {0.00/129 2} 80. Kg6 {+1.27/16 1} Kh8 {0.00/129 2} 81. Ra6
{+0.08/16 1} Kg8 {0.00/129 2} 82. Rf6 {+0.07/16 0} Kh8 {0.00/129 2} 83. Ra6
{+0.06/16 1} Kg8 {0.00/129 2} 84. Re6 {+0.05/17 1} Kh8 {0.00/129 2} 85. Rc6
{+0.04/16 1} Kg8 {0.00/129 2} 86. Ra6 {+0.05/18 1} ... {Black forfeits on
time} 1-0
[/pgn]

The final position has indeed occurred already 3 times, only the side to move is different.

[JVMerlino]

It seems there is a problem with repetition detection. Here is a sample game (played under Arena 3.5.1):

[pgn][Event "Myrddin 0.88 64-bit - Jul 19"]
[Site "Szentendre"]
[Date "2021.07.19"]
[Round "6"]
[White "Yace 0.99.87"]
[Black "Myrddin 0.88 64-bit"]
[Result "1-0"]
[ECO "D32"]
[Opening "QGD Tarrasch"]
[Time "19:46:20"]
[Variation "von Hennig-Schara Gambit, 7.Qxd5 Nc6 8.Bg5"]
[TimeControl "120+1"]
[PlyCount "171"]

1. d4 d5 2. c4 e6 3. Nc3 c5 4. cxd5 cxd4 5. Qa4+ {+0.74/10 8} Bd7 {-0.25/15
4} 6. Qxd4 {+0.65/11 7} exd5 {-0.39/14 3} 7. Qxd5 {+0.65/10 6} Nc6
{-0.34/13 3} 8. Bg5 {+0.50/10 5} Nf6 {-0.33/13 3} 9. Qd2 {+0.55/9 2} Be7
{-0.34/12 3} 10. Bxf6 {+0.50/9 2} Bxf6 {+0.13/13 3} 11. Rd1 {+0.28/9 4} Be6
{+0.07/13 3} 12. Qxd8+ {+0.10/9 3} Rxd8 {+0.49/15 3} 13. Rxd8+ {+0.07/10 3}
Kxd8 {+0.44/16 3} 14. a3 {+0.18/11 3} Kc7 {+0.44/14 3} 15. e4 {+0.36/9 1}
Re8 {+0.42/13 5} 16. Nge2 {+0.31/10 3} Bb3 {+0.32/12 3} 17. f4 {+0.53/11 3}
Rd8 {+0.41/12 3} 18. e5 {+0.48/11 3} Be7 {+0.36/13 3} 19. Nc1 {+0.49/10 3}
Be6 {+0.38/13 3} 20. Bd3 {+0.39/10 3} Nd4 {+0.29/12 4} 21. O-O {+0.67/10 3}
Nf5 {+0.35/12 2} 22. b4 {+0.81/10 2} h5 {+0.16/11 2} 23. Be4 {+0.87/10 2}
Kb8 {+0.13/12 2} 24. Re1 {+0.81/9 1} Rd4 {+0.03/12 2} 25. Nd3 {+0.82/9 2}
Rc4 {+0.12/14 2} 26. Na4 {+0.67/10 2} b6 {+0.35/14 3} 27. g3 {+0.64/10 2}
g6 {+0.50/13 2} 28. Ndb2 {+0.66/9 1} Rc8 {+0.48/14 2} 29. Nd1 {+0.43/10 2}
h4 {+0.56/14 2} 30. g4 {+0.68/10 1} Nd4 {+0.23/14 2} 31. Ne3 {+0.36/10 2}
Bd7 {+0.20/13 2} 32. Nb2 {+0.85/9 2} h3 {+0.08/13 2} 33. Rd1 {+0.89/10 2}
Ne2+ {+0.08/14 2} 34. Kf1 {+0.83/11 2} Bb5 {-0.04/15 2} 35. Bd3 {+0.92/11
2} Bxd3 {-0.39/16 6} 36. Nxd3 {+0.93/11 2} Nc3 {-0.39/16 2} 37. Rc1
{+0.92/11 2} a5 {-0.35/15 3} 38. g5 {+0.94/10 2} axb4 {-0.22/14 2} 39. Nxb4
{+0.88/10 1} Kb7 {-0.14/15 2} 40. Rc2 {+0.85/10 1} Ne4 {-0.45/15 4} 41. Rb2
{+0.83/10 2} Bc5 {+0.04/13 1} 42. Ned5 {+0.74/10 2} Bg1 {0.00/13 3} 43. Nd3
{+0.90/10 2} Rd8 {-0.02/13 1} 44. Kxg1 {+1.09/11 1} Rxd5 {-0.16/16 1} 45.
Nf2 {+0.98/11 1} Nxf2 {-0.26/17 1} 46. Kxf2 {+1.00/10 2} Rd3 {-0.25/17 2}
47. Re2 {+1.38/12 1} Rxa3 {-0.45/17 2} 48. e6 {+1.18/12 1} fxe6 {-0.44/17
1} 49. Rxe6 {+1.33/13 1} b5 {-0.34/16 2} 50. Rxg6 {+1.20/12 1} b4 {-0.34/16
2} 51. Re6 {+1.22/12 1} b3 {-0.21/15 2} 52. Re1 {+1.19/12 2} Kc7 {-0.30/15
1} 53. Rb1 {+1.13/12 2} Kd6 {-0.32/16 1} 54. Kf3 {+1.18/13 2} b2+ {-0.77/17
1} 55. Kg4 {+1.09/15 1} Rb3 {-0.85/19 1} 56. g6 {+1.28/15 1} Ke7 {-0.78/19
1} 57. Kg5 {+1.28/15 1} Rb4 {-0.32/20 7} 58. f5 {+1.25/14 1} Rb5 {-0.32/20
1} 59. g7 {+0.97/14 1} Kf7 {-0.32/20 1} 60. Kh6 {+0.92/14 1} Rb7 {-0.32/18
1} 61. f6 {+0.92/14 1} Rb4 {0.00/18 2} 62. Rg1 {+0.89/13 1} Rh4+ {0.00/17
1} 63. Kg5 Rb4 {0.00/17 1} 64. Kh5 {+0.89/13 1} Rb8 {0.00/16 1} 65. Rb1
{+0.89/14 1} Rb4 {0.00/19 1} 66. Kh6 {+0.83/13 1} Rh4+ {0.00/19 1} 67. Kg5
Rb4 {0.00/19 1} 68. Kf5 {+0.89/13 1} Rb8 {0.00/19 1} 69. Kf4 {+0.89/14 1}
Kxf6 {0.00/20 1} 70. Rxb2 {+0.98/14 1} Rg8 {0.00/23 1} 71. Rb3 {+0.98/14 0}
Rxg7 {0.00/27 1} 72. Rxh3 {+0.95/14 1} Rg8 {0.00/129 2} 73. Rh6+ {+1.01/12
1} Kf7 {0.00/129 2} 74. h4 {+1.11/13 1} Kf8 {0.00/129 2} 75. h5 {+1.16/13
1} Ke7 {0.00/129 2} 76. Ra6 {+1.25/13 1} Rh8 {0.00/129 2} 77. h6 {+1.43/14
1} Kf7 {0.00/129 2} 78. Kf5 {+1.40/15 1} Rd8 {0.00/129 2} 79. Rf6+
{+1.37/15 1} Kg8 {0.00/129 2} 80. Kg6 {+1.27/16 1} Kh8 {0.00/129 2} 81. Ra6
{+0.08/16 1} Kg8 {0.00/129 2} 82. Rf6 {+0.07/16 0} Kh8 {0.00/129 2} 83. Ra6
{+0.06/16 1} Kg8 {0.00/129 2} 84. Re6 {+0.05/17 1} Kh8 {0.00/129 2} 85. Rc6
{+0.04/16 1} Kg8 {0.00/129 2} 86. Ra6 {+0.05/18 1} ... {Black forfeits on
time} 1-0
[/pgn]

The final position has indeed occurred already 3 times, only the side to move is different.

Hello Gabor,

A threefold repetition can only be claimed if the same player is on the move in all three occurrences, and that player has had an identical list of legal moves in all three occurrences. So a threefold draw could not be claimed if the side to move is different.
As for draw detection in general, I can see that Myrddin had access to TBs, and was scoring the game as a draw starting with move 61. So I don't think there's a problem there.
But the main thing I'm concerned about is Myrddin losing on time. This never happens in my testing. Has it happened more than once? Does the Arena log file have some information about that?

Thanks very much for your testing!
jm

[Gabor Szots]

But the main thing I'm concerned about is Myrddin losing on time. This never happens in my testing. Has it happened more than once? Does the Arena log file have some information about that?

I regret to report that there were 19 or 20 time forfeits out of 512 games. Unfortunately no log is available. The only thing I can say is that mainbook and tourbook were not used (I deleted them from the Myrddin folder) but a general external book was used.

BTW, as I wrote upstream, this was under Arena where I usually have more time forfeits than other CCRL testers and some engines are particularly sensitive in this regard. Maybe my PC works on its limits.