Now playing around with Gull:
- ran a quick test game ( just to make sure I did not break anything ),
2m+2s - with 12 cores - no opening book - against the latest SF - the result might surprise you - this latest version of Gull is 15% faster on my mac pro - just with better PGO and other complier options - except to make the default number of Threads 1 and not 24 . Gull btw does not take kindly to HT (like most chess engines).
[pgn][Event "Computer Chess Game"]
[Site "Mac-Pro.local"]
[Date "2016.07.01"]
[Round "-"]
[White "Stockfish 062816-x"]
[Black "Gull 3 x64 (syzygy)"]
[Result "0-1"]
[TimeControl "120+2"]
[Annotator "1. +0.24 1... -0.15"]
1. e4 {+0.24/23} e5 {-0.15/17 2.4} 2. Nf3 {+0.27/22 5} Nc6 {-0.26/17 2.7}
3. Bb5 {+0.17/25 6} a6 {-0.26/18 3} 4. Ba4 {+0.12/25 9} Be7 {-0.08/16 2.9}
5. Bxc6 {+0.12/25 7} dxc6 {-0.03/17 1.5} 6. O-O {+0.12/23 5} Qd6
{+0.06/19 6} 7. b3 {+0.19/23 2.0} Bg4 {+0.13/19 6} 8. d3 {+0.06/24 7} O-O-O
{+0.11/18 6} 9. Be3 {+0.00/27 9} f5 {+0.38/17 1.6} 10. Nbd2 {-0.58/23 5} f4
{+0.28/18 2.2} 11. Nc4 {-0.76/21 0.7} Qe6 {+0.27/18 2.4} 12. Bd2
{-0.55/23 1.9} Bf6 {+0.34/19 6} 13. a4 {-0.46/26 18} Nh6 {+0.44/16 5} 14.
h3 {-0.44/23 10} Bh5 {+0.62/17 4} 15. b4 {-0.71/25 23} Nf7 {+0.60/17 4} 16.
Bc3 {-0.31/21 1.3} Qe7 {+0.72/18 13} 17. Rb1 {-0.34/21 4} Ng5 {+0.66/17 10}
18. Ncd2 {-0.41/21 3} b6 {+0.66/16 0.2} 19. a5 {-0.60/22 4} b5
{+0.70/15 1.3} 20. Kh1 {-0.91/27 15} Kb7 {+0.82/17 2.0} 21. Qe2
{-1.02/24 7} c5 {+0.84/17 2.3} 22. Rfc1 {-0.78/22 3} cxb4 {+0.99/17 1.7}
23. Bxb4 {-1.06/23 5} c5 {+0.97/17 2.3} 24. Ba3 {-1.09/18 1.2} Rhe8
{+1.05/17 2.7} 25. Kg1 {-1.07/19 2.7} Qc7 {+1.14/17 3} 26. Ra1 {-1.16/21 5}
Ne6 {+1.18/17 4} 27. Qe1 {-1.06/21 2.7} g5 {+1.34/16 1.2} 28. Nh2
{-1.02/25 1.9} Rg8 {+1.28/18 11} 29. f3 {-1.02/22 1.3} Bf7 {+1.27/19 6} 30.
Ng4 {-1.01/22 2.0} Bg7 {+1.27/18 0.1} 31. Nf2 {-1.18/23 2.7} h5
{+1.19/20 10} 32. Bb2 {-1.19/22 1.2} Nd4 {+1.25/21 9} 33. Nb1
{-1.07/21 2.8} Bf6 {+1.38/17 1.8} 34. Bxd4 {-1.24/21 2.0} exd4
{+1.70/18 1.2} 35. Nd2 {-1.02/25 2.0} Ka7 {+1.72/21 3} 36. Qe2
{-1.36/22 2.0} Rde8 {+1.72/23 4} 37. Rcb1 {-1.72/21 2.0} Rg6 {+1.81/20 2.8}
38. Nf1 {-1.70/21 1.1} Be6 {+1.81/21 3} 39. Nh2 {-1.78/24 2.9} Be5
{+1.81/22 3} 40. Qe1 {-1.88/23 2.0} Qd7 {+1.82/20 2.8} 41. Qc1
{-1.85/23 1.7} Rg7 {+1.82/19 3} 42. Rb2 {-1.98/23 2.3} Qc6 {+1.83/18 4} 43.
Rab1 {-2.06/23 2.0} Qc8 {+1.84/18 3} 44. Qe1 {-2.11/25 2.0} Rg6
{+1.84/19 4} 45. Ra1 {-2.01/24 2.0} Rh8 {+1.82/20 4} 46. Rbb1
{-2.22/23 2.0} Rhg8 {+1.82/22 3} 47. Qd1 {-2.26/23 2.0} Qd8 {+1.82/21 2.7}
48. Qc1 {-2.22/23 1.5} Qe7 {+1.82/21 3} 49. Qa3 {-2.31/23 2.5} R6g7
{+1.82/19 4} 50. Qc1 {-2.37/23 2.0} Re8 {+1.82/20 3} 51. Rb2 {-2.31/22 1.3}
Qd8 {+1.83/19 5} 52. Rab1 {-2.31/23 2.0} Qxa5 {+2.13/14 0.9} 53. Ra1
{-2.51/23 2.7} Qb6 {+2.08/16 1.3} 54. Qe1 {-2.67/25 2.0} Qc6 {+2.41/19 4}
55. Rbb1 {-2.72/20 2.0} Bc7 {+2.69/17 5} 56. Qd1 {-3.13/20 2.0} a5
{+2.97/16 1.9} 57. e5 {-3.12/19 2.0} Bf5 {+3.36/16 3} 58. Nf1
{-3.34/21 2.0} Rxe5 {+3.30/15 1.0} 59. Rb2 {-3.64/21 2.0} a4 {+3.40/17 2.3}
60. Rab1 {-3.89/22 2.0} b4 {+3.82/17 1.2} 61. Ra1 {-4.30/22 2.0} a3
{+3.99/17 2.3} 62. Nh2 {-4.26/23 1.4} Re3 {+4.06/16 4} 63. Qd2
{-4.33/23 2.6} Qa6 {+4.06/17 4} 64. Rb3 {-4.65/20 2.0} Be6 {+4.22/16 2.4}
65. Rbb1 {-4.34/20 0.9} Bd5 {+4.16/17 2.2} 66. Rf1 {-4.67/22 3} Qe6
{+4.50/16 1.2} 67. Qd1 {-5.29/20 2.0} Bd6 {+4.45/18 4} 68. Kh1
{-5.48/23 2.0} g4 {+4.53/18 3} 69. hxg4 {-5.20/21 1.0} hxg4 {+4.83/17 1.2}
70. Nhxg4 {-5.94/23 3} Qe7 {+5.76/15 1.1} 71. Nh2 {-7.21/20 2.0} Qh4
{+6.38/15 0.7} 72. Nh3 {-7.88/23 2.0} Be6 {+6.86/16 1.1} 73. Rb1
{-8.75/23 2.0} Bxh3 {+7.82/17 1.2} 74. gxh3 {-10.48/23 2.0} Rge7
{+8.46/17 2.6} 75. Rg1 {-16.34/24 2.0} Re2 {+10.83/17 0.9} 76. Qf1
{-17.69/25 2.0} Rf2 {+11.48/19 1.3} 77. Rg4 {-19.70/23 2.0} Rxf1+
{+13.20/18 19} 78. Rxf1 {-19.82/22 1.1} Qxh3 {+13.20/17 2.0} 79. Rg2
{-17.60/24 1.2} Rh7 {+15.05/16 2.0} 80. Rgf2 {-24.72/24 3} Be7
{+20.64/13 0.6} 81. c4 {-1000.10/47 2.4} dxc3 {+22.81/13 0.6} 82. d4
{-1000.09/48 0.2} b3 {+1000.14/14 1.7} 83. Ra1 {-1000.08/37 1.5} cxd4
{+1000.08/15 0.6} 84. Kg1 {-1000.07/49 1.6} d3 {+1000.06/17 0.6} 85. Re1
{-1000.06/54 1.6} d2 {+1000.05/20 1.8} 86. Ra1 {-1000.05/63 0.3} Bc5
{+1000.05/22 2.5} 87. Kh1 {-1000.04/127 0.3} c2 {+1000.04/28 1.7} 88. Rxd2
{-1000.03/127 0.1} Qxf3+ {+1000.03/31 2.0} 89. Rg2 {-1000.02/1 0.1} Rxh2+
{+1000.02/31 5} 90. Kxh2 {-1000.01/1 0.1} Qh5# {+1000.02/31 5}
{Black mates} 0-1[/pgn]
Gull 3 x64
Moderators: hgm, Rebel, chrisw
-
- Posts: 1339
- Joined: Fri Nov 02, 2012 9:43 am
- Location: New Delhi, India
Re: Gull 3 x64
abrok.eu shows that latest Gull version was released in February.
Is this the version you are using ?
Is this the version you are using ?
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: Gull 3 x64
No , I'm using a version maintained by Basil Falcinelli on Github.MikeB wrote:Now playing around with Gull:
- ran a quick test game ( just to make sure I did not break anything ),
2m+2s - with 12 cores - no opening book - against the latest SF - the result might surprise you - this latest version of Gull is 15% faster on my mac pro - just with better PGO and other complier options - except to make the default number of Threads 1 and not 24 . Gull btw does not take kindly to HT (like most chess engines).
[pgn][Event "Computer Chess Game"]
[Site "Mac-Pro.local"]
[Date "2016.07.01"]
[Round "-"]
[White "Stockfish 062816-x"]
[Black "Gull 3 x64 (syzygy)"]
[Result "0-1"]
[TimeControl "120+2"]
[Annotator "1. +0.24 1... -0.15"]
1. e4 {+0.24/23} e5 {-0.15/17 2.4} 2. Nf3 {+0.27/22 5} Nc6 {-0.26/17 2.7}
3. Bb5 {+0.17/25 6} a6 {-0.26/18 3} 4. Ba4 {+0.12/25 9} Be7 {-0.08/16 2.9}
5. Bxc6 {+0.12/25 7} dxc6 {-0.03/17 1.5} 6. O-O {+0.12/23 5} Qd6
{+0.06/19 6} 7. b3 {+0.19/23 2.0} Bg4 {+0.13/19 6} 8. d3 {+0.06/24 7} O-O-O
{+0.11/18 6} 9. Be3 {+0.00/27 9} f5 {+0.38/17 1.6} 10. Nbd2 {-0.58/23 5} f4
{+0.28/18 2.2} 11. Nc4 {-0.76/21 0.7} Qe6 {+0.27/18 2.4} 12. Bd2
{-0.55/23 1.9} Bf6 {+0.34/19 6} 13. a4 {-0.46/26 18} Nh6 {+0.44/16 5} 14.
h3 {-0.44/23 10} Bh5 {+0.62/17 4} 15. b4 {-0.71/25 23} Nf7 {+0.60/17 4} 16.
Bc3 {-0.31/21 1.3} Qe7 {+0.72/18 13} 17. Rb1 {-0.34/21 4} Ng5 {+0.66/17 10}
18. Ncd2 {-0.41/21 3} b6 {+0.66/16 0.2} 19. a5 {-0.60/22 4} b5
{+0.70/15 1.3} 20. Kh1 {-0.91/27 15} Kb7 {+0.82/17 2.0} 21. Qe2
{-1.02/24 7} c5 {+0.84/17 2.3} 22. Rfc1 {-0.78/22 3} cxb4 {+0.99/17 1.7}
23. Bxb4 {-1.06/23 5} c5 {+0.97/17 2.3} 24. Ba3 {-1.09/18 1.2} Rhe8
{+1.05/17 2.7} 25. Kg1 {-1.07/19 2.7} Qc7 {+1.14/17 3} 26. Ra1 {-1.16/21 5}
Ne6 {+1.18/17 4} 27. Qe1 {-1.06/21 2.7} g5 {+1.34/16 1.2} 28. Nh2
{-1.02/25 1.9} Rg8 {+1.28/18 11} 29. f3 {-1.02/22 1.3} Bf7 {+1.27/19 6} 30.
Ng4 {-1.01/22 2.0} Bg7 {+1.27/18 0.1} 31. Nf2 {-1.18/23 2.7} h5
{+1.19/20 10} 32. Bb2 {-1.19/22 1.2} Nd4 {+1.25/21 9} 33. Nb1
{-1.07/21 2.8} Bf6 {+1.38/17 1.8} 34. Bxd4 {-1.24/21 2.0} exd4
{+1.70/18 1.2} 35. Nd2 {-1.02/25 2.0} Ka7 {+1.72/21 3} 36. Qe2
{-1.36/22 2.0} Rde8 {+1.72/23 4} 37. Rcb1 {-1.72/21 2.0} Rg6 {+1.81/20 2.8}
38. Nf1 {-1.70/21 1.1} Be6 {+1.81/21 3} 39. Nh2 {-1.78/24 2.9} Be5
{+1.81/22 3} 40. Qe1 {-1.88/23 2.0} Qd7 {+1.82/20 2.8} 41. Qc1
{-1.85/23 1.7} Rg7 {+1.82/19 3} 42. Rb2 {-1.98/23 2.3} Qc6 {+1.83/18 4} 43.
Rab1 {-2.06/23 2.0} Qc8 {+1.84/18 3} 44. Qe1 {-2.11/25 2.0} Rg6
{+1.84/19 4} 45. Ra1 {-2.01/24 2.0} Rh8 {+1.82/20 4} 46. Rbb1
{-2.22/23 2.0} Rhg8 {+1.82/22 3} 47. Qd1 {-2.26/23 2.0} Qd8 {+1.82/21 2.7}
48. Qc1 {-2.22/23 1.5} Qe7 {+1.82/21 3} 49. Qa3 {-2.31/23 2.5} R6g7
{+1.82/19 4} 50. Qc1 {-2.37/23 2.0} Re8 {+1.82/20 3} 51. Rb2 {-2.31/22 1.3}
Qd8 {+1.83/19 5} 52. Rab1 {-2.31/23 2.0} Qxa5 {+2.13/14 0.9} 53. Ra1
{-2.51/23 2.7} Qb6 {+2.08/16 1.3} 54. Qe1 {-2.67/25 2.0} Qc6 {+2.41/19 4}
55. Rbb1 {-2.72/20 2.0} Bc7 {+2.69/17 5} 56. Qd1 {-3.13/20 2.0} a5
{+2.97/16 1.9} 57. e5 {-3.12/19 2.0} Bf5 {+3.36/16 3} 58. Nf1
{-3.34/21 2.0} Rxe5 {+3.30/15 1.0} 59. Rb2 {-3.64/21 2.0} a4 {+3.40/17 2.3}
60. Rab1 {-3.89/22 2.0} b4 {+3.82/17 1.2} 61. Ra1 {-4.30/22 2.0} a3
{+3.99/17 2.3} 62. Nh2 {-4.26/23 1.4} Re3 {+4.06/16 4} 63. Qd2
{-4.33/23 2.6} Qa6 {+4.06/17 4} 64. Rb3 {-4.65/20 2.0} Be6 {+4.22/16 2.4}
65. Rbb1 {-4.34/20 0.9} Bd5 {+4.16/17 2.2} 66. Rf1 {-4.67/22 3} Qe6
{+4.50/16 1.2} 67. Qd1 {-5.29/20 2.0} Bd6 {+4.45/18 4} 68. Kh1
{-5.48/23 2.0} g4 {+4.53/18 3} 69. hxg4 {-5.20/21 1.0} hxg4 {+4.83/17 1.2}
70. Nhxg4 {-5.94/23 3} Qe7 {+5.76/15 1.1} 71. Nh2 {-7.21/20 2.0} Qh4
{+6.38/15 0.7} 72. Nh3 {-7.88/23 2.0} Be6 {+6.86/16 1.1} 73. Rb1
{-8.75/23 2.0} Bxh3 {+7.82/17 1.2} 74. gxh3 {-10.48/23 2.0} Rge7
{+8.46/17 2.6} 75. Rg1 {-16.34/24 2.0} Re2 {+10.83/17 0.9} 76. Qf1
{-17.69/25 2.0} Rf2 {+11.48/19 1.3} 77. Rg4 {-19.70/23 2.0} Rxf1+
{+13.20/18 19} 78. Rxf1 {-19.82/22 1.1} Qxh3 {+13.20/17 2.0} 79. Rg2
{-17.60/24 1.2} Rh7 {+15.05/16 2.0} 80. Rgf2 {-24.72/24 3} Be7
{+20.64/13 0.6} 81. c4 {-1000.10/47 2.4} dxc3 {+22.81/13 0.6} 82. d4
{-1000.09/48 0.2} b3 {+1000.14/14 1.7} 83. Ra1 {-1000.08/37 1.5} cxd4
{+1000.08/15 0.6} 84. Kg1 {-1000.07/49 1.6} d3 {+1000.06/17 0.6} 85. Re1
{-1000.06/54 1.6} d2 {+1000.05/20 1.8} 86. Ra1 {-1000.05/63 0.3} Bc5
{+1000.05/22 2.5} 87. Kh1 {-1000.04/127 0.3} c2 {+1000.04/28 1.7} 88. Rxd2
{-1000.03/127 0.1} Qxf3+ {+1000.03/31 2.0} 89. Rg2 {-1000.02/1 0.1} Rxh2+
{+1000.02/31 5} 90. Kxh2 {-1000.01/1 0.1} Qh5# {+1000.02/31 5}
{Black mates} 0-1[/pgn]
https://github.com/basil00/Gull , and his most recent updates were about a month ago.
I like his forked version for two reasons, he has maintained compatiabilty with the mac computers and also has implemented syzygy table base support. You can follow the latest version of Hawkeye (new name for my gull derivative) running in this test tournament here:
http://talkchess.com/forum/viewtopic.php?t=60669
or if you just want to see the results or the pgn files, you many find them here:
Results: https://www.dropbox.com/s/cx9vntj8n8myw ... a.txt?dl=1
PGN Files:https://www.dropbox.com/s/tmwepshusdy48h7/all.pgn?dl=1
It is currently private , no source or exe's have been released. Although I do plan to release the exe and source at some point in the future. One comment about the source , if you us the flag -Wall, you will see all of the compiler warnings, Complier warnings should be treated as if the compiler is telling you - "hey I can do what you are telling me to do, but are are you really sure you want me to do that - I think you are making a mistake "
Except for some comestic changes I made to Gull's output , most of the ELO gain in Gull can be attributed to fixing "just some - maybe 1/3" of the compiler warnings and some speed optimizations. Some of the warnings are hard to track down since the original source was so obtuse. Basil has done a good job with some of his comments.
-
- Posts: 1339
- Joined: Fri Nov 02, 2012 9:43 am
- Location: New Delhi, India
Re: Gull 3 x64
Hi
How to get the Gull UCI/exe Engine ?
How to get the Gull UCI/exe Engine ?
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis