Introducing Toga III chess engine

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Introducing Toga III chess engine

Post by dkappe »

Toga rides again. Most of you will be familiar with my work creating NNUE nets from non-stockfish engines. One of the first of those was Toga II 4.0.2. I thought it only appropriate that these nets run in the original engine.

Introducing Toga III 0.2, with the latest Toga III 0.5 net embedded. The only change from the Toga II 4.0.2 code base is that the eval has been replaced with the nnue code from Daniel Shawul’s nnue-probe library.

Right now the search doesn’t do incremental nnue updates, so is losing about 20% nps. I have plans to fix that as well as add support for Syzygy tablebases.

https://github.com/dkappe/TogaIII

If anyone is able to contribute Linux and especially Windows builds, that would be most appreciated.

[pgn]
[Event "?"]
[Site "?"]
[Date "2021.01.10"]
[Round "3"]
[White "toga3.2"]
[Black "arasan"]
[Result "1-0"]
[ECO "C69"]
[GameDuration "00:03:02"]
[GameEndTime "2021-01-10T10:23:00.941 CST"]
[GameStartTime "2021-01-10T10:19:58.049 CST"]
[Opening "Ruy Lopez"]
[PlyCount "124"]
[Termination "adjudication"]
[TimeControl "60+1"]
[Variation "Exchange Variation , 5.O-O"]

1. e4 {book} e5 {book} 2. Nf3 {book} Nc6 {book} 3. Bb5 {book} a6 {book}
4. Bxc6 {book} dxc6 {book} 5. O-O {book} Ne7 {book} 6. Nxe5 {book} Qd4 {book}
7. Qh5 {book} g6 {book} 8. Qg5 {book} Bg7 {book} 9. Nd3 {book} f5 {book}
10. e5 {book} Qg4 {-0.08/23 4.2s} 11. Qe3 {+0.52/17 1.9s} Nd5 {-0.46/24 8.0s}
12. Qe1 {+0.51/18 1.8s} f4 {-0.49/23 2.6s} 13. f3 {+0.38/17 1.9s}
Qf5 {-0.65/23 3.9s} 14. Nc3 {+0.50/15 1.8s} a5 {-0.35/22 2.7s}
15. Nxd5 {+0.31/14 1.7s} cxd5 {-0.45/22 3.6s} 16. b4 {+0.26/15 2.9s}
O-O {-0.62/20 3.1s} 17. Bb2 {+0.16/15 1.6s} Bd7 {-0.50/21 2.3s}
18. Qf2 {+0.31/14 2.2s} Bb5 {-0.71/22 4.4s} 19. Qd4 {+0.42/16 2.3s}
Bc4 {-1.16/20 2.6s} 20. Rfe1 {+0.35/15 2.7s} axb4 {-1.24/22 3.9s}
21. Nxb4 {+0.41/15 2.0s} Ra5 {-1.29/24 2.9s} 22. Qc3 {+1.17/17 2.1s}
Qe6 {-1.31/24 2.2s} 23. d3 {+0.99/16 1.6s} Ra4 {-1.49/21 3.8s}
24. a3 {+1.19/16 1.9s} Ba6 {-1.50/23 1.9s} 25. Qb3 {+1.03/15 2.4s}
Ra5 {-1.29/21 2.5s} 26. Rab1 {+1.02/14 1.8s} b6 {-0.99/21 2.3s}
27. Nxa6 {+1.03/15 1.8s} Rxa6 {-1.14/20 2.0s} 28. c4 {+1.21/16 2.3s}
dxc4 {-1.01/22 1.6s} 29. dxc4 {+1.17/15 2.2s} Re8 {-1.02/21 3.5s}
30. Re4 {+1.35/15 1.6s} g5 {-1.39/22 1.7s} 31. h4 {+1.27/16 2.2s}
h6 {-1.47/21 1.5s} 32. Rd1 {+1.12/15 3.6s} Bf8 {-1.57/21 2.0s}
33. Kh2 {+1.13/14 1.4s} Raa8 {-1.43/22 1.5s} 34. a4 {+1.11/15 2.7s}
Bc5 {-1.04/19 1.4s} 35. Bd4 {+1.17/15 2.2s} Be7 {-1.01/18 1.9s}
36. Bf2 {+1.26/13 1.3s} Red8 {-1.39/17 1.4s} 37. Rd5 {+1.31/13 1.5s}
Ra7 {-1.42/17 1.4s} 38. Bc5 {+1.48/13 1.9s} bxc5 {-3.67/19 4.8s}
39. Rxd8+ {+2.50/13 0.49s} Bxd8 {-3.91/24 1.6s} 40. Qb8 {+2.61/14 0.76s}
Rxa4 {-3.98/22 1.2s} 41. Qxd8+ {+2.80/14 0.98s} Kh7 {-4.31/21 1.2s}
42. hxg5 {+2.93/15 2.0s} hxg5 {-4.10/24 1.4s} 43. Qxg5 {+3.36/13 1.2s}
Qh6+ {-4.43/22 1.3s} 44. Qxh6+ {+3.57/14 0.70s} Kxh6 {-4.42/26 1.2s}
45. e6 {+3.85/17 1.9s} Ra8 {-4.63/25 1.9s} 46. Rxf4 {+4.31/15 1.4s}
Kg6 {-4.85/25 1.1s} 47. g4 {+4.58/16 1.1s} Re8 {-5.41/23 1.3s}
48. Re4 {+4.80/16 0.37s} Kf6 {-7.49/23 1.1s} 49. f4 {+5.66/17 1.2s}
Kg6 {-10.01/25 1.1s} 50. f5+ {+6.46/19 1.2s} Kg5 {-13.58/26 1.1s}
51. e7 {+7.90/22 1.1s} Kf6 {-20.46/26 1.1s} 52. Kg3 {+9.98/27 1.2s}
Rxe7 {-23.11/27 1.1s} 53. Rxe7 {+12.23/28 1.1s} Kxe7 {-22.12/25 1.1s}
54. g5 {+12.71/20 0.52s} Kf8 {-240.18/25 1.1s} 55. Kg4 {+14.28/26 1.4s}
Ke8 {-M28/25 1.1s} 56. f6 {+19.22/41 1.2s} Kf7 {-M24/24 0.28s}
57. Kf5 {+20.93/24 0.38s} Kf8 {-M22/22 0.092s} 58. g6 {+22.21/23 0.38s}
Kg8 {-M20/20 0.016s} 59. Ke6 {+M87/26 1.3s} Kf8 {-M18/18 0.022s}
60. g7+ {+M85/63 0.005s} Kg8 {-M16/16 0.002s} 61. Ke7 {+M83/63 0.005s}
c6 {-M14/14 0.001s} 62. f7+ {+M59/63 0.006s}
Kxg7 {-M10/19 0.41s, White wins by adjudication: SyzygyTB} 1-0
[/pgn]
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Top
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Introducing Toga III chess engine

Post by MikeB »

dkappe wrote: ↑Sun Jan 10, 2021 5:39 pm Toga rides again. Most of you will be familiar with my work creating NNUE nets from non-stockfish engines. One of the first of those was Toga II 4.0.2. I thought it only appropriate that these nets run in the original engine.

Introducing Toga III 0.2, with the latest Toga III 0.5 net embedded. The only change from the Toga II 4.0.2 code base is that the eval has been replaced with the nnue code from Daniel Shawul’s nnue-probe library.

Right now the search doesn’t do incremental nnue updates, so is losing about 20% nps. I have plans to fix that as well as add support for Syzygy tablebases.

https://github.com/dkappe/TogaIII

If anyone is able to contribute Linux and especially Windows builds, that would be most appreciated.

[pgn]
[Event "?"]
[Site "?"]
[Date "2021.01.10"]
[Round "3"]
[White "toga3.2"]
[Black "arasan"]
[Result "1-0"]
[ECO "C69"]
[GameDuration "00:03:02"]
[GameEndTime "2021-01-10T10:23:00.941 CST"]
[GameStartTime "2021-01-10T10:19:58.049 CST"]
[Opening "Ruy Lopez"]
[PlyCount "124"]
[Termination "adjudication"]
[TimeControl "60+1"]
[Variation "Exchange Variation , 5.O-O"]

1. e4 {book} e5 {book} 2. Nf3 {book} Nc6 {book} 3. Bb5 {book} a6 {book}
4. Bxc6 {book} dxc6 {book} 5. O-O {book} Ne7 {book} 6. Nxe5 {book} Qd4 {book}
7. Qh5 {book} g6 {book} 8. Qg5 {book} Bg7 {book} 9. Nd3 {book} f5 {book}
10. e5 {book} Qg4 {-0.08/23 4.2s} 11. Qe3 {+0.52/17 1.9s} Nd5 {-0.46/24 8.0s}
12. Qe1 {+0.51/18 1.8s} f4 {-0.49/23 2.6s} 13. f3 {+0.38/17 1.9s}
Qf5 {-0.65/23 3.9s} 14. Nc3 {+0.50/15 1.8s} a5 {-0.35/22 2.7s}
15. Nxd5 {+0.31/14 1.7s} cxd5 {-0.45/22 3.6s} 16. b4 {+0.26/15 2.9s}
O-O {-0.62/20 3.1s} 17. Bb2 {+0.16/15 1.6s} Bd7 {-0.50/21 2.3s}
18. Qf2 {+0.31/14 2.2s} Bb5 {-0.71/22 4.4s} 19. Qd4 {+0.42/16 2.3s}
Bc4 {-1.16/20 2.6s} 20. Rfe1 {+0.35/15 2.7s} axb4 {-1.24/22 3.9s}
21. Nxb4 {+0.41/15 2.0s} Ra5 {-1.29/24 2.9s} 22. Qc3 {+1.17/17 2.1s}
Qe6 {-1.31/24 2.2s} 23. d3 {+0.99/16 1.6s} Ra4 {-1.49/21 3.8s}
24. a3 {+1.19/16 1.9s} Ba6 {-1.50/23 1.9s} 25. Qb3 {+1.03/15 2.4s}
Ra5 {-1.29/21 2.5s} 26. Rab1 {+1.02/14 1.8s} b6 {-0.99/21 2.3s}
27. Nxa6 {+1.03/15 1.8s} Rxa6 {-1.14/20 2.0s} 28. c4 {+1.21/16 2.3s}
dxc4 {-1.01/22 1.6s} 29. dxc4 {+1.17/15 2.2s} Re8 {-1.02/21 3.5s}
30. Re4 {+1.35/15 1.6s} g5 {-1.39/22 1.7s} 31. h4 {+1.27/16 2.2s}
h6 {-1.47/21 1.5s} 32. Rd1 {+1.12/15 3.6s} Bf8 {-1.57/21 2.0s}
33. Kh2 {+1.13/14 1.4s} Raa8 {-1.43/22 1.5s} 34. a4 {+1.11/15 2.7s}
Bc5 {-1.04/19 1.4s} 35. Bd4 {+1.17/15 2.2s} Be7 {-1.01/18 1.9s}
36. Bf2 {+1.26/13 1.3s} Red8 {-1.39/17 1.4s} 37. Rd5 {+1.31/13 1.5s}
Ra7 {-1.42/17 1.4s} 38. Bc5 {+1.48/13 1.9s} bxc5 {-3.67/19 4.8s}
39. Rxd8+ {+2.50/13 0.49s} Bxd8 {-3.91/24 1.6s} 40. Qb8 {+2.61/14 0.76s}
Rxa4 {-3.98/22 1.2s} 41. Qxd8+ {+2.80/14 0.98s} Kh7 {-4.31/21 1.2s}
42. hxg5 {+2.93/15 2.0s} hxg5 {-4.10/24 1.4s} 43. Qxg5 {+3.36/13 1.2s}
Qh6+ {-4.43/22 1.3s} 44. Qxh6+ {+3.57/14 0.70s} Kxh6 {-4.42/26 1.2s}
45. e6 {+3.85/17 1.9s} Ra8 {-4.63/25 1.9s} 46. Rxf4 {+4.31/15 1.4s}
Kg6 {-4.85/25 1.1s} 47. g4 {+4.58/16 1.1s} Re8 {-5.41/23 1.3s}
48. Re4 {+4.80/16 0.37s} Kf6 {-7.49/23 1.1s} 49. f4 {+5.66/17 1.2s}
Kg6 {-10.01/25 1.1s} 50. f5+ {+6.46/19 1.2s} Kg5 {-13.58/26 1.1s}
51. e7 {+7.90/22 1.1s} Kf6 {-20.46/26 1.1s} 52. Kg3 {+9.98/27 1.2s}
Rxe7 {-23.11/27 1.1s} 53. Rxe7 {+12.23/28 1.1s} Kxe7 {-22.12/25 1.1s}
54. g5 {+12.71/20 0.52s} Kf8 {-240.18/25 1.1s} 55. Kg4 {+14.28/26 1.4s}
Ke8 {-M28/25 1.1s} 56. f6 {+19.22/41 1.2s} Kf7 {-M24/24 0.28s}
57. Kf5 {+20.93/24 0.38s} Kf8 {-M22/22 0.092s} 58. g6 {+22.21/23 0.38s}
Kg8 {-M20/20 0.016s} 59. Ke6 {+M87/26 1.3s} Kf8 {-M18/18 0.022s}
60. g7+ {+M85/63 0.005s} Kg8 {-M16/16 0.002s} 61. Ke7 {+M83/63 0.005s}
c6 {-M14/14 0.001s} 62. f7+ {+M59/63 0.006s}
Kxg7 {-M10/19 0.41s, White wins by adjudication: SyzygyTB} 1-0
[/pgn]
Sweet , I will make an avx2 build and also add it to Harmon for Android. Thank you!
Image
Top
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Introducing Toga III chess engine

Post by dkappe »

I’ve made a release with avx2, sse and vanilla Linux binaries.

https://github.com/dkappe/TogaIII/relea ... /Toga3-0.2
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Top
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Introducing Toga III chess engine

Post by dkappe »

Wasp 4.5 seems very strong.

[pgn]

[Event "?"]
[Site "?"]
[Date "2021.01.10"]
[Round "88"]
[White "toga3.2"]
[Black "wasp4.5"]
[Result "1-0"]
[ECO "E12"]
[GameDuration "00:03:35"]
[GameEndTime "2021-01-10T13:48:42.174 CST"]
[GameStartTime "2021-01-10T13:45:06.939 CST"]
[Opening "Queen's Indian"]
[PlyCount "153"]
[TimeControl "60+1"]
[Variation "4.Nc3"]

1. d4 {book} Nf6 {book} 2. c4 {book} e6 {book} 3. Nf3 {book} b6 {book}
4. Nc3 {book} Bb4 {book} 5. Bg5 {book} Bb7 {book} 6. Nd2 {book}
O-O {+0.09/17 2.3s} 7. a3 {-0.23/14 2.1s} Be7 {+0.14/20 4.7s}
8. e3 {-0.25/15 2.0s} c5 {+0.28/18 3.1s} 9. dxc5 {-0.30/15 1.9s}
bxc5 {+0.26/18 3.1s} 10. Nf3 {-0.28/15 2.0s} d6 {+0.30/18 3.1s}
11. Qc2 {-0.17/16 2.0s} h6 {+0.24/17 3.0s} 12. Bh4 {-0.17/17 2.3s}
Nbd7 {+0.45/17 2.9s} 13. Be2 {-0.14/17 2.2s} g5 {+0.56/18 2.1s}
14. Bg3 {+0.09/14 0.52s} g4 {+0.04/18 2.3s} 15. Nd2 {+0.31/16 1.9s}
Bxg2 {+0.18/18 2.0s} 16. Rg1 {+0.25/15 1.0s} Bc6 {-0.22/19 3.7s}
17. O-O-O {+0.52/16 2.5s} Kh8 {0.00/19 2.5s} 18. Nde4 {+0.48/16 2.5s}
Nxe4 {-0.45/21 8.1s} 19. Nxe4 {+0.59/14 0.67s} f5 {-0.43/18 2.2s}
20. Nxd6 {+0.63/16 2.5s} Rb8 {-0.17/17 2.1s} 21. Nb5 {+0.47/15 2.5s}
Rb7 {0.00/16 2.0s} 22. Bf4 {+0.54/14 1.9s} Be4 {0.00/18 1.5s}
23. Qd2 {+0.31/15 1.5s} Qb6 {-0.57/18 5.2s} 24. Bc7 {+0.47/14 2.6s}
Rxc7 {+1.12/19 1.4s} 25. Nxc7 {+0.22/13 0.58s} Rd8 {+0.79/19 1.6s}
26. Nb5 {0.00/16 1.5s} a6 {+0.87/19 1.3s} 27. Nc3 {-0.04/18 3.1s}
Ne5 {+0.93/19 1.8s} 28. Qe1 {0.00/16 1.8s} Rb8 {+0.74/18 1.3s}
29. Na4 {0.00/18 2.2s} Qc7 {+0.72/18 1.7s} 30. Qd2 {-0.17/16 2.0s}
h5 {+0.65/17 1.7s} 31. Nc3 {+0.26/16 1.7s} Bc6 {+0.71/16 1.2s}
32. e4 {+0.46/14 2.4s} f4 {+0.50/17 1.2s} 33. Kb1 {+0.60/14 2.3s}
Qb7 {+0.71/17 1.8s} 34. Ka1 {+0.34/14 2.1s} Bg5 {+0.22/17 2.1s}
35. f3 {+0.97/15 1.3s} h4 {+0.23/16 1.1s} 36. Rxg4 {+0.82/14 1.5s}
Nxg4 {-1.10/18 1.6s} 37. Rg1 {+0.66/15 0.50s} Rd8 {-1.21/18 1.5s}
38. Qc1 {+0.42/15 0.46s} Qe7 {-0.89/18 1.1s} 39. Rxg4 {+1.21/14 0.51s}
Kh7 {-1.20/18 1.4s} 40. Bd1 {+1.23/15 1.6s} Kh6 {-1.02/17 1.1s}
41. Bc2 {+1.16/15 2.0s} Be8 {-1.22/17 1.1s} 42. e5 {+0.87/17 2.2s}
Bh5 {-0.53/21 1.2s} 43. Rg1 {+1.26/20 2.1s} Bxf3 {-0.52/19 1.1s}
44. Qf1 {+1.02/19 1.7s} Bc6 {-0.48/19 1.1s} 45. Be4 {+1.29/18 1.4s}
Be8 {-0.97/19 1.4s} 46. Qh3 {+1.17/16 2.1s} Rd4 {-1.03/18 1.0s}
47. Bb1 {+1.86/17 1.6s} Bc6 {-1.50/20 1.4s} 48. Ne2 {+2.83/18 1.3s}
f3 {-3.58/19 1.8s} 49. Nxd4 {+3.29/16 1.2s} cxd4 {-4.78/19 1.1s}
50. Qg4 {+3.57/17 1.3s} a5 {-5.04/19 1.3s} 51. Qxd4 {+3.81/17 1.4s}
Qf7 {-5.70/19 1.3s} 52. Qg4 {+4.35/19 1.7s} Qe7 {-5.94/21 0.97s}
53. Bc2 {+4.59/19 1.7s} a4 {-5.90/19 1.3s} 54. Bd1 {+4.83/17 1.3s}
Qg7 {-6.69/19 0.96s} 55. Qxe6+ {+5.35/17 1.1s} Qg6 {-6.99/19 0.96s}
56. Qc8 {+5.57/17 1.2s} Kg7 {-7.69/17 0.97s} 57. Qc7+ {+5.92/16 1.7s}
Kh6 {-8.13/19 0.97s} 58. Qb8 {+5.99/13 0.35s} Be8 {-9.26/19 2.0s}
59. Bxf3 {+6.32/13 0.41s} Bf7 {-12.50/17 1.7s} 60. Qh8+ {+11.02/18 0.39s}
Qh7 {-12.02/12 0.018s} 61. Qf8+ {+10.91/15 0.46s} Qg7 {-18.52/22 1.2s}
62. Qxg7+ {+11.23/15 0.40s} Kxg7 {-17.64/15 0.018s} 63. Rxg5+ {+11.35/16 0.42s}
Kh6 {-19.96/23 0.97s} 64. Rg4 {+12.04/15 0.41s} Be6 {-23.91/21 1.6s}
65. c5 {+13.12/17 0.44s} h3 {-27.34/20 1.7s} 66. Rd4 {+16.28/28 1.5s}
Bc8 {-32.07/20 1.5s} 67. Bg4 {+18.61/25 0.61s} Bb7 {-46.51/19 2.9s}
68. c6 {+20.69/23 1.1s} Ba8 {-M24/16 0.98s} 69. c7 {+23.57/29 1.3s}
Bb7 {-M18/15 0.81s} 70. Rb4 {+25.99/24 0.40s} Kg5 {-M16/15 0.19s}
71. Rxb7 {+297.44/33 1.9s} Kxg4 {-M14/13 0.017s} 72. c8=Q+ {+M115/63 0.032s}
Kf4 {-M12/11 0.007s} 73. e6 {+M9/24 1.2s} Ke4 {-M10/9 0.047s}
74. e7 {+M7/63 0.26s} Kf4 {-M8/7 0.005s} 75. e8=Q {+M5/63 0.033s}
Kf3 {-M6/5 0.003s} 76. Qxh3+ {+M3/63 0.011s} Kf2 {-M4/3 0.003s}
77. Qee3# {+M1/63 0.008s, White mates} 1-0
[/pgn]
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Top
Damir
Posts: 2801
Joined: Mon Feb 11, 2008 3:53 pm
Location: Denmark
Full name: Damir Desevac

Re: Introducing Toga III chess engine

Post by Damir »

Hi Dietrich

Any Windows binary ? My system not support Linux....
Top
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Introducing Toga III chess engine

Post by dkappe »

Damir wrote: ↑Sun Jan 10, 2021 8:53 pm Hi Dietrich

Any Windows binary ? My system not support Linux....
Unfortunately I don’t do Windows, but I have a few requests out.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Top
Angle
Posts: 319
Joined: Sat Oct 31, 2020 1:04 am
Full name: Aleksey Glebov

Re: Introducing Toga III chess engine

Post by Angle »

Hi, Dietrich!

Nice news!
I am also interested in Windows binaries :P

Today I have installed Toga III 0.1 with 0.4 net via the NetRunner (under Windows 10). I have launched a quick testrun against the selected engines. At the moment two mini-matches are played with TC 20s + 3s:
Toga III 0.1 - Gull 3 : 14,5 - 5,5 (+9 -0 =11) 72,5% +173
Toga III 0.1 - Xiphos 0.6 : 13,0 - 7,0 (+7 -1 =12) 65% +108

The other 3 opponents will be Ethereal 12.75, Komodo 11.01, and Stockfish 10.

I am going to include both Toga III v.0.1 with 0.4 net and Toga III v.0.2 with 0.5 net to my planned two strong tournaments. The first of them will start in 7-10 days.

The surprising thing concerning Toga and NetRunner is that the fastest compiles of NR for me are popcnt and SSE3 while avx2 and bmi2 are approx. 2 times(!) slower. So I use the popcnt version, though I guess something is wrong here :roll: as my system supports all modern sets of instructions.

Hmm, Ethereal was destroyed with the same result as Xiphos. Just amazing!
Toga III 0.1 - Ethereal 12.75 : 13,0 - 7,0 (+7 -1 =12) 65% +108
Incredibly fast systems miscount incredibly fast.
Top
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Introducing Toga III chess engine

Post by dkappe »

I added an avx2 Windows build with the shared libnnueprobe.so included. Works under wine, but I have no other way of testing.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Top
dkappe
Posts: 1631
Joined: Tue Aug 21, 2018 7:52 pm
Full name: Dietrich Kappe

Re: Introducing Toga III chess engine

Post by dkappe »

Angle wrote: ↑Sun Jan 10, 2021 9:47 pm Hi, Dietrich!

Nice news!
I am also interested in Windows binaries :P

Today I have installed Toga III 0.1 with 0.4 net via the NetRunner (under Windows 10). I have launched a quick testrun against the selected engines. At the moment two mini-matches are played with TC 20s + 3s:
Toga III 0.1 - Gull 3 : 14,5 - 5,5 (+9 -0 =11) 72,5% +173
Toga III 0.1 - Xiphos 0.6 : 13,0 - 7,0 (+7 -1 =12) 65% +108

The other 3 opponents will be Ethereal 12.75, Komodo 11.01, and Stockfish 10.

I am going to include both Toga III v.0.1 with 0.4 net and Toga III v.0.2 with 0.5 net to my planned two strong tournaments. The first of them will start in 7-10 days.

The surprising thing concerning Toga and NetRunner is that the fastest compiles of NR for me are popcnt and SSE3 while avx2 and bmi2 are approx. 2 times(!) slower. So I use the popcnt version, though I guess something is wrong here :roll: as my system supports all modern sets of instructions.

Hmm, Ethereal was destroyed with the same result as Xiphos. Just amazing!
Toga III 0.1 - Ethereal 12.75 : 13,0 - 7,0 (+7 -1 =12) 65% +108
With NetRunner you are getting the benefit of the highly efficient SF search. Running in the Toga II search, it’s maybe 90 elo under Wasp 4.5 and 50 elo under the latest Arasan.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Top
User avatar
Werner
Posts: 2871
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Re: Introducing Toga III chess engine

Post by Werner »

dkappe wrote: ↑Sun Jan 10, 2021 11:01 pm I added an avx2 Windows build with the shared libnnueprobe.so included. Works under wine, but I have no other way of testing.
Hi Dietrich, I got it working under Windows with Arena 3.5.1.
Thanks for the .exe
best wishes
Werner
Werner
Top

Return to “Computer Chess Club: General Topics”