Twin 2080Tis and large NN

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

Moderators: hgm, Rebel, chrisw

shrapnel
Posts: 1339
Joined: Fri Nov 02, 2012 9:43 am
Location: New Delhi, India

Re: Twin 2080Tis and large NN

Post by shrapnel »

corres wrote: Fri Oct 25, 2019 9:32 pmOpening positions was the same in both matches.
Leela parameters were also the same in the case of the two net.
So enhancing the move time (or the power of GPUs) the result tend to reach equality.
It would be interesting to make a match with Time = 1 min + 200 sec / move (~ FIDE tempo)
This is the mistake the Testers at lc0.org are making, at least as far as Competitive Chess against AB Engines is concerned.
While your Tests are very interesting and have their own significance, there is a BIG difference between playing different NNs of lc0 against each other and playing lc0 against powerful AB Engines like Stockfish and its derivatives.
For example, if I use one of the smaller NNs like the 128x10 or 256x20 on my System, lc0 gets "smoked" nearly every time against even relatively weak CPUs running Stockfish, no matter the very high Knps or Time Control used.
Its ONLY using the very large 384x30 Net that enables lc0 to match upto the AB Engine, whatever the TC. Of course, a strong GPU or two is required to fully utilize the large NN, unless one wants to run it at 1 Knps or so, which is again pointless.
So, in the Real World (in a manner of speaking) of competitive Engine-Engine Matches a very large NN is required, supported by a strong GPU or two, at least at the present stage of lc0 development, no matter what the Theorists say.
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis
corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: Twin 2080Tis and large NN

Post by corres »

shrapnel wrote: Sat Oct 26, 2019 7:24 am
corres wrote: Fri Oct 25, 2019 9:32 pmOpening positions was the same in both matches.
Leela parameters were also the same in the case of the two net.
So enhancing the move time (or the power of GPUs) the result tend to reach equality.
It would be interesting to make a match with Time = 1 min + 200 sec / move (~ FIDE tempo)
This is the mistake the Testers at lc0.org are making, at least as far as Competitive Chess against AB Engines is concerned.
While your Tests are very interesting and have their own significance, there is a BIG difference between playing different NNs of lc0 against each other and playing lc0 against powerful AB Engines like Stockfish and its derivatives.
For example, if I use one of the smaller NNs like the 128x10 or 256x20 on my System, lc0 gets "smoked" nearly every time against even relatively weak CPUs running Stockfish, no matter the very high Knps or Time Control used.
I think if you would use a special opening book what is appropriate to the stile of NN engines and it utilizes the weakness of AB engines you could get better results. Maybe the "Attacker Book" is not the best for an NN engine.
shrapnel wrote: Sat Oct 26, 2019 7:24 am Its ONLY using the very large 384x30 Net that enables lc0 to match upto the AB Engine, whatever the TC. Of course, a strong GPU or two is required to fully utilize the large NN, unless one wants to run it at 1 Knps or so, which is again pointless.
So, in the Real World (in a manner of speaking) of competitive Engine-Engine Matches a very large NN is required, supported by a strong GPU or two, at least at the present stage of lc0 development, no matter what the Theorists say.
In general a good big NN is stronger than a small one and it needs one or more stronger GPU(s)
to realize its full potential.
But TCEC and CCC proved there are strong mid range 256x20 nets also. It is pity they need
3-4 x RTX 2080 TI against Stockfish running on powerful PC.
Btw.
What a kind of engine is "Eman"?
As I know it is a weaker derivative of Stockfish - like Sugar, CorChess, etc.
Leo
Posts: 1080
Joined: Fri Sep 16, 2016 6:55 pm
Location: USA/Minnesota
Full name: Leo Anger

Re: Twin 2080Tis and large NN

Post by Leo »

mwyoung wrote: Fri Oct 25, 2019 7:20 pm
shrapnel wrote: Fri Oct 25, 2019 6:11 pm Large NNs (384x30) really do improve the performance of lc0( whether Laskos Sir fully agrees or not ). Of course, twin 2080 Tis help :) , but on the other hand using a large NN with a weak GPU makes no sense and would simply backfire.
Following is a Game against a guy using 16 Cores and Stockfish..

[pgn][Event "Rated game, 16 min"] [Site "Engine Room"] [Date "2019.10.25"] [Round "?"] [White "Mrtambourineman, Stockfish 260819 64"] [Black "Shrapnel, Lc0 v0.22.0_1"] [Result "1/2-1/2"] [ECO "E11"] [WhiteElo "2424"] [Annotator "0.00;0.15"] [PlyCount "123"] [EventDate "2019.10.25"] [SourceTitle "playchess.com"] [TimeControl "960"] {Stockfish 260819 64 POPCNTpp (16 threads): 41.9 plies; 23,493kN/s AMD Ryzen 7 2700X Eight-Core Processor 3693MHz, (8 cores, 16 threads), Attaker Pro 3. ctg, 8192 MB} 1. d4 {[%eval 0,0] [%emt 0:00:00]} Nf6 {[%eval 0,0] [%emt 0:00: 00]} 2. c4 {[%eval 0,0] [%emt 0:00:00]} e6 {[%eval 0,0] [%emt 0:00:00]} 3. Nf3 {[%eval 0,0] [%emt 0:00:00]} d5 {[%eval 0,0] [%emt 0:00:00]} 4. g3 {[%eval 0,0] [%emt 0:00:00]} Bb4+ {[%eval 0,0] [%emt 0:00:00]} 5. Bd2 {[%eval 0,0] [%emt 0: 00:00]} Be7 {[%eval 0,0] [%emt 0:00:00]} 6. Bg2 {[%eval 0,0] [%emt 0:00:00]} O-O {[%eval 0,0] [%emt 0:00:00]} 7. O-O {[%eval 0,0] [%emt 0:00:00]} c6 { [%eval 0,0] [%emt 0:00:00]} 8. Qc2 {[%eval 0,0] [%emt 0:00:00]} Nbd7 {[%eval 0, 0] [%emt 0:00:00]} 9. Rc1 {[%eval 0,0] [%emt 0:00:00]} b6 {[%eval 0,0] [%emt 0: 00:00]} 10. cxd5 {[%eval 0,0] [%emt 0:00:00]} cxd5 {[%eval 0,0] [%emt 0:00:00]} 11. Qb3 {[%eval 0,0] [%emt 0:00:00]} Ne4 {[%eval 0,0] [%emt 0:00:00]} 12. Bb4 { [%eval 0,0] [%emt 0:00:00]} Bxb4 {[%eval 0,0] [%emt 0:00:00]} 13. Qxb4 { [%eval 0,0] [%emt 0:00:00]} a5 {[%eval 0,0] [%emt 0:00:00]} 14. Qa3 {[%eval 0, 0] [%emt 0:00:00]} Ba6 {[%eval 0,0] [%emt 0:00:00]} 15. e3 {[%eval 0,0] [%emt 0:00:00]} Qb8 {[%eval 0,0] [%emt 0:00:00]} 16. Nbd2 {[%eval 0,0] [%emt 0:00:00] } Nxd2 {[%eval 0,0] [%emt 0:00:00]} 17. Nxd2 {[%eval 0,0] [%emt 0:00:00]} Ra7 { [%eval 0,0] [%emt 0:00:00]} 18. Rc3 {[%eval 0,0] [%emt 0:00:00]} Be2 {[%eval 0, 0] [%emt 0:00:00]} 19. Bf1 {[%eval 0,0] [%emt 0:00:00]} b5 {[%eval 0,0] [%emt 0:00:00]} 20. Rc2 {[%eval 0,0] [%emt 0:00:00]} Bc4 {[%eval 0,0] [%emt 0:00:00]} 21. b3 {[%eval 0,0] [%emt 0:00:00]} Bxf1 {[%eval 0,0] [%emt 0:00:00]} 22. Kxf1 {[%eval 0,0] [%emt 0:00:00]} a4 {[%eval 0,0] [%emt 0:00:00]} 23. Kg2 {[%eval 0, 0] [%emt 0:00:00]} Qa8 {[%eval 0,0] [%emt 0:00:00]} 24. b4 {[%eval 0,0] [%emt 0:00:00]} Nb6 {[%eval 0,0] [%emt 0:00:00]} 25. Rc5 {[%eval 0,0] [%emt 0:00:00]} Rb7 {[%eval 0,0] [%emt 0:00:00]} 26. Qd3 {[%eval 0,0] [%emt 0:00:00]} Nd7 { [%eval 0,0] [%emt 0:00:00]} 27. Rc6 {[%eval 0,0] [%emt 0:00:00]} Nb6 {[%eval 0, 0] [%emt 0:00:00]} 28. Nb1 {[%eval 0,0] [%emt 0:00:00]} Rbb8 {[%eval 0,0] [%emt 0:00:00]} 29. Rc5 {[%eval 0,0] [%emt 0:00:00]} Nc4 {[%eval 0,0] [%emt 0: 00:00]} 30. Nc3 {[%eval 0,0] [%emt 0:00:00]} Nd6 {[%eval 0,0] [%emt 0:00:00]} 31. f3 {[%eval 0,0] [%emt 0:00:00]} h6 {[%eval 15,9] [%emt 0:00:24]} 32. Rc1 { [%eval 0,42] [%emt 0:01:18] (h4)} Rfc8 {[%eval 3,12] [%emt 0:00:22] (Tfd8)} 33. Rd1 {[%eval 0,39] [%emt 0:01:42] (e4)} Nc4 {[%eval -3,12] [%emt 0:00:30] (Db7)} 34. Rc1 {[%eval 0,40] [%emt 0:01:00] (Rb1)} a3 {[%eval -14,12] [%emt 0:00:46] (Da6)} 35. Rb1 {[%eval 0,39] [%emt 0:00:11] (h4)} Nb2 {[%eval -18,12] [%emt 0: 00:47]} 36. Qe2 {[%eval 0,44] [%emt 0:00:00] (Qd2)} Rxc5 {[%eval -24,13] [%emt 0:00:51] (Db7)} 37. dxc5 {[%eval 27,29] [%emt 0:00:16]} Qc6 {[%eval -22,14] [%emt 0:00:45]} 38. Rf1 {[%eval 0,36] [%emt 0:00:47] (h4)} e5 {[%eval -23,13] [%emt 0:01:22] (Cc4)} 39. e4 {[%eval 0,32] [%emt 0:00:19]} d4 {[%eval -20,13] [%emt 0:00:00]} 40. Nd5 {[%eval 0,34] [%emt 0:00:17]} Kh7 {[%eval -18,12] [%emt 0:00:52] (Tb7)} 41. f4 {[%eval 0,38] [%emt 0:00:34] (g4)} exf4 {[%eval -58,23] [%emt 0:00:34]} 42. Rxf4 {[%eval 0,42] [%emt 0:00:00]} Qe6 {[%eval -57, 13] [%emt 0:00:00] (d3)} 43. Qf3 {[%eval 74,35] [%emt 0:00:17]} f6 {[%eval 53, 22] [%emt 0:01:17]} 44. Qxa3 {[%eval 31,39] [%emt 0:00:00]} Nc4 {[%eval 54,24] [%emt 0:00:00]} 45. Qd3 {[%eval 47,37] [%emt 0:00:18]} Ra8 {[%eval 54,16] [%emt 0:00:00]} 46. Rf2 {[%eval 25,38] [%emt 0:00:36]} Qe5 {[%eval 53,20] [%emt 0:00:00]} 47. Qf3 {[%eval 32,40] [%emt 0:00:42]} Kh8 {[%eval 53,20] [%emt 0:00:21] (Ta6)} 48. Qf4 {[%eval 58,32] [%emt 0:00:19] (Kg1)} d3 {[%eval 46,21] [%emt 0:01:31]} 49. Qxe5 {[%eval 46,50] [%emt 0:00:00]} Nxe5 {[%eval 46, 20] [%emt 0:00:02]} 50. Nc3 {[%eval 35,43] [%emt 0:00:14]} Nc4 {[%eval 46,23] [%emt 0:00:01]} 51. Kf1 {[%eval 35,43] [%emt 0:00:08] (a4)} Ra3 {[%eval 42,20] [%emt 0:00:56]} 52. Nb1 {[%eval 13,48] [%emt 0:00:31] (Nd5)} Ra4 {[%eval 39,25] [%emt 0:00:38]} 53. a3 {[%eval 2,49] [%emt 0:00:00]} Nxa3 {[%eval 38,25] [%emt 0:00:14]} 54. Nxa3 {[%eval 2,50] [%emt 0:00:00]} Rxa3 {[%eval 38,24] [%emt 0: 00:10]} 55. Ke1 {[%eval 2,51] [%emt 0:00:00]} Kg8 {[%eval 38,22] [%emt 0:00:12] } 56. Kd2 {[%eval 2,50] [%emt 0:00:06]} Kf7 {[%eval 38,21] [%emt 0:00:08]} 57. c6 {[%eval 0,41] [%emt 0:00:01] (e5)} Ke6 {[%eval 17,16] [%emt 0:01:25]} 58. Rf5 {[%eval 0,54] [%emt 0:00:00]} Kd6 {[%eval 17,16] [%emt 0:00:00]} 59. Rxb5 { [%eval 0,50] [%emt 0:00:07]} Kxc6 {[%eval 16,14] [%emt 0:00:00]} 60. Rc5+ { [%eval 0,49] [%emt 0:00:16]} Kd6 {[%eval 15,13] [%emt 0:00:14]} 61. Rd5+ { [%eval 0,52] [%emt 0:00:00]} Kc7 {[%eval 15,12] [%emt 0:00:17] (Rc6)} 62. Ra5 { [%eval 0,44] [%emt 0:00:09] (Rc5+)} 1/2-1/2[/pgn]

Against a slightly weaker opponent, lc0 won, even though it was a 3 minute Game, supposedly not lc0's forte....

[pgn][Event "Rated game, 3 min"] [Site "Engine Room"] [Date "2019.10.25"] [Round "?"] [White "Shrapnel, Lc0 v0.22.0"] [Black "JML26, Stockfish 260819 64"] [Result "1-0"] [ECO "A28"] [WhiteElo "2479"] [BlackElo "2422"] [Annotator "0.57;-0.09"] [PlyCount "145"] [EventDate "2019.10.25"] [SourceTitle "playchess.com"] [TimeControl "180"] {Stockfish 260819 64 (6 cores): 28.5 plies; 11,387kN/s Intel(R) Core(TM) i5-8400 CPU @ 2.80GHz 2807MHz, Attaker Pro 3.ctg, 8192 MB} 1. c4 {[%eval 0,0] [%emt 0:00:00]} e5 {[%eval -9,24] [%emt 0:00:04]} 2. Nc3 {[%eval 0,0] [%emt 0: 00:00]} Nf6 {[%eval -13,27] [%emt 0:00:09]} 3. Nf3 {[%eval 0,0] [%emt 0:00:00]} Nc6 {[%eval -22,24] [%emt 0:00:02]} 4. e4 {[%eval 0,0] [%emt 0:00:00] (a2-a3)} Bb4 {[%eval -38,25] [%emt 0:00:03]} 5. d3 {[%eval 0,0] [%emt 0:00:00]} d6 { [%eval -14,28] [%emt 0:00:11]} 6. a3 {[%eval 0,0] [%emt 0:00:00]} Bc5 {[%eval 0,27] [%emt 0:00:03]} 7. b4 {[%eval 0,0] [%emt 0:00:00]} Bb6 {[%eval 0,0] [%emt 0:00:00]} 8. Be3 {[%eval 0,0] [%emt 0:00:00]} Bg4 {[%eval -13,26] [%emt 0:00:05]} 9. Be2 {[%eval 0,0] [%emt 0:00:00]} O-O {[%eval 0,28] [%emt 0:00:06]} 10. O-O {[%eval 0,0] [%emt 0:00:00]} Bxf3 {[%eval 0,28] [%emt 0:00:01]} 11. Bxf3 {[%eval 0,0] [%emt 0:00:00]} Nd4 {[%eval 0,28] [%emt 0:00:01]} 12. Na4 { [%eval 0,0] [%emt 0:00:00]} Ne6 {[%eval 0,27] [%emt 0:00:02]} 13. Nxb6 { [%eval 0,0] [%emt 0:00:00]} axb6 {[%eval 4,26] [%emt 0:00:01]} 14. Qd2 { [%eval 0,0] [%emt 0:00:00]} h6 {[%eval 30,28] [%emt 0:00:10]} 15. a4 {[%eval 0, 0] [%emt 0:00:00]} c5 {[%eval 29,24] [%emt 0:00:03]} 16. Bd1 {[%eval 57,5] [%emt 0:00:05] (Tf1-b1)} Nd7 {[%eval 24,23] [%emt 0:00:03]} 17. Rb1 {[%eval 47, 9] [%emt 0:00:05] (g2-g3)} Nb8 {[%eval 4,26] [%emt 0:00:04] (Nd4)} 18. b5 { [%eval 43,10] [%emt 0:00:01]} Nd7 {[%eval 25,26] [%emt 0:00:02]} 19. Qe2 { [%eval 52,9] [%emt 0:00:03] (Tb1-a1)} Qh4 {[%eval 8,24] [%emt 0:00:03]} 20. Ra1 {[%eval 51,11] [%emt 0:00:08]} Nd4 {[%eval 0,33] [%emt 0:00:00] (Nf4)} 21. Qg4 {[%eval 57,12] [%emt 0:00:04]} Qxg4 {[%eval 0,36] [%emt 0:00:00]} 22. Bxg4 { [%eval 55,13] [%emt 0:00:02]} f5 {[%eval 0,29] [%emt 0:00:04] (Nf6)} 23. exf5 { [%eval 64,16] [%emt 0:00:05]} Nc2 {[%eval 0,36] [%emt 0:00:00]} 24. Ra2 { [%eval 53,17] [%emt 0:00:07]} Nb4 {[%eval 0,38] [%emt 0:00:00]} 25. Ra3 { [%eval 55,20] [%emt 0:00:03] (Ta2-a1)} Nc2 {[%eval 0,37] [%emt 0:00:02]} 26. Rc3 {[%eval 55,22] [%emt 0:00:02] (Ta3-a2)} Nd4 {[%eval 0,35] [%emt 0:00:02]} 27. Ra3 {[%eval 55,21] [%emt 0:00:01]} Nc2 {[%eval 0,38] [%emt 0:00:03]} 28. Ra2 {[%eval 55,21] [%emt 0:00:01] (Ta3-c3)} Nb4 {[%eval 11,39] [%emt 0:00:20]} 29. Raa1 {[%eval 56,23] [%emt 0:00:00]} Nc2 {[%eval 13,28] [%emt 0:00:03]} 30. Bf3 {[%eval 57,22] [%emt 0:00:00] (Ta1-c1)} Nxa1 {[%eval 0,26] [%emt 0:00:02]} 31. Rxa1 {[%eval 56,20] [%emt 0:00:00]} Ra7 {[%eval -10,27] [%emt 0:00:02]} 32. g4 {[%eval 56,20] [%emt 0:00:00]} Kf7 {[%eval 0,28] [%emt 0:00:02] (Rfa8)} 33. h4 {[%eval 77,14] [%emt 0:00:22]} Rfa8 {[%eval 0,38] [%emt 0:00:00]} 34. Bd1 { [%eval 84,13] [%emt 0:00:12]} Nf6 {[%eval 0,38] [%emt 0:00:00]} 35. Bd2 { [%eval 93,15] [%emt 0:00:07]} Rf8 {[%eval 46,31] [%emt 0:00:07] (Ng8)} 36. Bf3 {[%eval 140,11] [%emt 0:00:09] (Rg1-g2)} Ke7 {[%eval 0,31] [%emt 0:00:02]} 37. Kg2 {[%eval 132,13] [%emt 0:00:02] (g4-g5)} Ng8 {[%eval 0,32] [%emt 0:00:02]} 38. Kg3 {[%eval 128,13] [%emt 0:00:07] (Ff3-e4)} Kd7 {[%eval 15,30] [%emt 0:00: 02]} 39. Kh3 {[%eval 129,13] [%emt 0:00:02] (Ff3-e4)} Ke7 {[%eval 0,30] [%emt 0:00:05]} 40. Bh1 {[%eval 146,15] [%emt 0:00:11] (Rh3-g3)} Kd7 {[%eval 70,28] [%emt 0:00:07]} 41. f4 {[%eval 152,18] [%emt 0:00:00]} exf4 {[%eval 65,27] [%emt 0:00:01]} 42. Bc3 {[%eval 161,17] [%emt 0:00:03] (Fh1-e4)} Nf6 {[%eval 80,25] [%emt 0:00:02]} 43. Bf3 {[%eval 162,13] [%emt 0:00:00]} Re8 {[%eval 111, 28] [%emt 0:00:04]} 44. Bd2 {[%eval 165,15] [%emt 0:00:01]} Kc7 {[%eval 114,30] [%emt 0:00:00]} 45. Bxf4 {[%eval 165,14] [%emt 0:00:02]} Rh8 {[%eval 110,29] [%emt 0:00:00] (Raa8)} 46. Kg3 {[%eval 168,12] [%emt 0:00:10]} Raa8 {[%eval 119,34] [%emt 0:00:00]} 47. Bd2 {[%eval 169,13] [%emt 0:00:01] (g4-g5)} Nd7 { [%eval 98,22] [%emt 0:00:01]} 48. Bc3 {[%eval 175,13] [%emt 0:00:04]} Ne5 { [%eval 201,27] [%emt 0:00:01]} 49. Be4 {[%eval 178,14] [%emt 0:00:00] (Ff3-d5)} Rhf8 {[%eval 106,24] [%emt 0:00:02]} 50. h5 {[%eval 194,11] [%emt 0:00:06] (a4-a5)} Rf7 {[%eval 94,23] [%emt 0:00:03]} 51. Kf4 {[%eval 199,12] [%emt 0:00: 03]} Rh8 {[%eval 101,27] [%emt 0:00:02] (Re7)} 52. a5 {[%eval 218,11] [%emt 0: 00:05]} bxa5 {[%eval 152,29] [%emt 0:00:00]} 53. Bxa5+ {[%eval 215,12] [%emt 0: 00:05] (Ta1xa5)} Kb8 {[%eval 157,23] [%emt 0:00:01]} 54. Bc3 {[%eval 216,11] [%emt 0:00:00]} Nd7 {[%eval 177,25] [%emt 0:00:03]} 55. g5 {[%eval 220,11] [%emt 0:00:01]} hxg5+ {[%eval 205,25] [%emt 0:00:00]} 56. Kxg5 {[%eval 218,10] [%emt 0:00:01]} Ne5 {[%eval 206,25] [%emt 0:00:00] (Re7)} 57. b6 {[%eval 241,8] [%emt 0:00:03]} Re7 {[%eval 230,27] [%emt 0:00:01] (Rhf8)} 58. Rh1 {[%eval 247, 9] [%emt 0:00:02]} Rh6 {[%eval 260,28] [%emt 0:00:01]} 59. Bd5 {[%eval 251,10] [%emt 0:00:01] (Th1-f1)} Re8 {[%eval 222,20] [%emt 0:00:00]} 60. Rh3 {[%eval 245,10] [%emt 0:00:03] (d3-d4)} Re7 {[%eval 327,26] [%emt 0:00:02] (Reh8)} 61. d4 {[%eval 258,8] [%emt 0:00:01] (Th3-h1)} cxd4 {[%eval 278,22] [%emt 0:00:00]} 62. Bxd4 {[%eval 236,9] [%emt 0:00:00]} Re8 {[%eval 368,23] [%emt 0:00:00]} 63. Rh1 {[%eval 232,8] [%emt 0:00:01]} Rf6 {[%eval 438,27] [%emt 0:00:02] (Reh8)} 64. Bc3 {[%eval 237,8] [%emt 0:00:00] (Th1-f1)} Ref8 {[%eval 294,20] [%emt 0: 00:00] (Rh6)} 65. Rf1 {[%eval 234,9] [%emt 0:00:00] (Fd5-e6)} Re8 {[%eval 442, 25] [%emt 0:00:01] (Rh6)} 66. Bb2 {[%eval 239,7] [%emt 0:00:01] (Fc3-d4)} Kc8 { [%eval 400,20] [%emt 0:00:01]} 67. Ra1 {[%eval 247,9] [%emt 0:00:00]} Kb8 { [%eval 507,26] [%emt 0:00:00]} 68. Ra7 {[%eval 278,10] [%emt 0:00:00]} Re7 { [%eval 658,25] [%emt 0:00:00]} 69. c5 {[%eval 338,9] [%emt 0:00:00] (Fb2-d4)} Nf7+ {[%eval 619,21] [%emt 0:00:00]} 70. Kf4 {[%eval 368,9] [%emt 0:00:00]} Ne5 {[%eval 930,25] [%emt 0:00:01]} 71. c6 {[%eval 465,8] [%emt 0:00:00]} Nxc6 { [%eval 1208,24] [%emt 0:00:00] (Nd3+)} 72. Bxf6 {[%eval 879,7] [%emt 0:00:00]} Nb4 {[%eval 1773,26] [%emt 0:00:01] (gxf6)} 73. Bxe7 {[%eval 4891,6] [%emt 0: 00:00] JML26,Stockfish 260819 64 abandonne (Lag: Av=0.47s, max=1.0s)} 1-0[/pgn]
Thanks for the games. It looks like the best CPU you played was a older 8 core 2700x CPU, and this game was a draw. I have not seen any data that suggest using multi GPUs with Lc0 is worth the investment at this point. But this could change.... Right now SF is just smoking Lc0 and the other NN in my testing.

Stockfish and a good CPU provides the best chess value and strength.
Do you know why SF lost to LCO in the superfinal 2 seasons ago? How did it win the latest TCEC?
Advanced Micro Devices fan.
shrapnel
Posts: 1339
Joined: Fri Nov 02, 2012 9:43 am
Location: New Delhi, India

Re: Twin 2080Tis and large NN

Post by shrapnel »

corres wrote: Sat Oct 26, 2019 1:10 pmMaybe the "Attacker Book" is not the best for an NN engine.
Agreed. Can you point out a Book more suitable for lc0 ?
In fact, I don't like using an Opening Book for lc0, but unfortunately the AB Engines slaughter it if it doesn't use one.
So, no choice.
corres wrote: Sat Oct 26, 2019 1:10 pmWhat kind of engine is "Eman"?
I don't know if its weaker than Stockfish, but it's a very popular Engine being used many Players on Playchess and InfinityChess, so it must be quite good.
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis
corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: Twin 2080Tis and large NN

Post by corres »

shrapnel wrote: Sun Oct 27, 2019 4:10 am
corres wrote: Sat Oct 26, 2019 1:10 pmMaybe the "Attacker Book" is not the best for an NN engine.
Agreed. Can you point out a Book more suitable for lc0 ?
In fact, I don't like using an Opening Book for lc0, but unfortunately the AB Engines slaughter it if it doesn't use one.
So, no choice.
[/quote]

I also do not know special book for NN engines. But I think because of the similarity between playing stile of humans and NN engines a book made from well selected GM games may be appropriate to NN engines too. Another opportunity: You collect the games of NN engines (won and drawn) from TCEC and CCC and other site and you make from them a book.
In general it is a good strategy to use closed openings to NN engines. Stockfis is too strong in opened and half opened games.
corres
Posts: 3657
Joined: Wed Nov 18, 2015 11:41 am
Location: hungary

Re: Twin 2080Tis and large NN

Post by corres »

shrapnel wrote: Sun Oct 27, 2019 4:10 am
corres wrote: Sat Oct 26, 2019 1:10 pm Maybe the "Attacker Book" is not the best for an NN engine.
Agreed. Can you point out a Book more suitable for lc0 ?
In fact, I don't like using an Opening Book for lc0, but unfortunately the AB Engines slaughter it if it doesn't use one.
So, no choice.
I also do not know special book for NN engines. But I think because of the similarity between playing stile of humans and NN engines a book made from well selected GM games maybe appropriate to NN engines too. Another opportunity: You collect the games of NN engines (won and drawn) from TCEC and CCC and other site and you make from them a book.
In general it is a good strategy to use closed openings with NN engines. Stockfis is too strong in opened and half opened games.
Hai
Posts: 598
Joined: Sun Aug 04, 2013 1:19 pm

Re: Twin 2080Tis and large NN

Post by Hai »

shrapnel wrote: Fri Oct 25, 2019 6:11 pm Large NNs (384x30) really do improve the performance of lc0( whether Laskos Sir fully agrees or not ). Of course, twin 2080 Tis help :) , but on the other hand using a large NN with a weak GPU makes no sense and would simply backfire.
That's right.
I'm using 30x384 and 2x RTX 2080 Ti and it's great:)
LC0 is playing much stronger and finds more key-moves, than before. The strength is increasing much more with more time.
User avatar
Nordlandia
Posts: 2821
Joined: Fri Sep 25, 2015 9:38 pm
Location: Sortland, Norway

Re: Twin 2080Tis and large NN

Post by Nordlandia »

On paper 2080 Ti SLi is a monster.
shrapnel
Posts: 1339
Joined: Fri Nov 02, 2012 9:43 am
Location: New Delhi, India

Re: Twin 2080Tis and large NN

Post by shrapnel »

Hai wrote: Tue Oct 29, 2019 9:22 amThat's right.
I'm using 30x384 and 2x RTX 2080 Ti and it's great:)
LC0 is playing much stronger and finds more key-moves, than before. The strength is increasing much more with more time.
Yes, indeed. The Sergei 30x384 nets are better than anything that the official lco.org has produced. They're simply wasting time and resources on 128x24 nets, though I've been told that they are simply a means to an end and ultimately very strong nets will be produced.
In the meantime, Sergei's NNs will do very nicely for me :) .
Nordlandia wrote: Tue Oct 29, 2019 10:22 am On paper 2080 Ti SLi is a monster.
On paper ? Get a load of this Game against a 36 Core Xeon Monster using Stockfish
[pgn][Event "Rated game, 5 min"] [Site "Engine Room"] [Date "2019.10.29"] [Round "?"] [White "Shrapnel, Lc0 v0.22.0"] [Black "McGudna, Stockfish 140719 64"] [Result "1-0"] [ECO "B85"] [WhiteElo "2429"] [BlackElo "2477"] [Annotator "-0.15;-0.39"] [PlyCount "201"] [EventDate "2019.10.29"] [SourceTitle "playchess.com"] [TimeControl "300"] {Stockfish 140719 64 BMI2 - POLYGLOT (36 cores): 47.6 plies; 52,342kN/s Intel (R) Xeon(R) CPU E5-2696 v3 @ 2.30GHz 2295MHz, (36 cores, 72 threads), Attaker Pro 3.ctg, 8192 MB} 1. e4 {[%eval 0,0] [%emt 0:00:00]} c5 {[%eval 0,0] [%emt 0: 00:00]} 2. Nf3 {[%eval 0,0] [%emt 0:00:00]} d6 {[%eval 0,0] [%emt 0:00:00]} 3. Nc3 {[%eval 0,0] [%emt 0:00:00]} a6 {[%eval 0,0] [%emt 0:00:00]} 4. d4 { [%eval 0,0] [%emt 0:00:00]} cxd4 {[%eval 0,0] [%emt 0:00:00]} 5. Nxd4 {[%eval 0,0] [%emt 0:00:00]} Nf6 {[%eval 0,0] [%emt 0:00:00]} 6. Be2 {[%eval 0,0] [%emt 0:00:00]} e6 {[%eval 0,0] [%emt 0:00:00]} 7. Be3 {[%eval 0,0] [%emt 0:00: 00]} Nc6 {[%eval 0,0] [%emt 0:00:00]} 8. O-O {[%eval 0,0] [%emt 0:00:00]} Be7 { [%eval 0,0] [%emt 0:00:00]} 9. f4 {[%eval 0,0] [%emt 0:00:00]} O-O {[%eval 0,0] [%emt 0:00:00]} 10. Kh1 {[%eval 0,0] [%emt 0:00:00]} Qc7 {[%eval 0,0] [%emt 0: 00:00]} 11. Qe1 {[%eval 0,0] [%emt 0:00:00]} Nxd4 {[%eval 0,0] [%emt 0:00:00]} 12. Bxd4 {[%eval 0,0] [%emt 0:00:00]} b5 {[%eval 0,0] [%emt 0:00:00]} 13. a3 { [%eval 0,0] [%emt 0:00:00]} Nd7 {[%eval 0,0] [%emt 0:00:00]} 14. Qg3 {[%eval 0, 0] [%emt 0:00:00]} Bf6 {[%eval 0,0] [%emt 0:00:00]} 15. Bxf6 {[%eval 0,0] [%emt 0:00:00]} Nxf6 {[%eval 0,0] [%emt 0:00:00]} 16. Rad1 {[%eval 0,0] [%emt 0:00:00]} Bb7 {[%eval -39,28] [%emt 0:00:09]} 17. Qe3 {[%eval 0,0] [%emt 0:00: 00] (Td4)} Rfe8 {[%eval -61,28] [%emt 0:00:12]} 18. Bf3 {[%eval 0,0] [%emt 0: 00:00]} h6 {[%eval -57,29] [%emt 0:00:08]} 19. Rd3 {[%eval -15,9] [%emt 0:00: 07] (Tfe1)} e5 {[%eval -52,30] [%emt 0:00:06]} 20. Rfd1 {[%eval -13,15] [%emt 0:00:02]} exf4 {[%eval -35,34] [%emt 0:00:13] (Qc5)} 21. Qxf4 {[%eval -8,12] [%emt 0:00:07]} Re6 {[%eval -37,36] [%emt 0:00:04]} 22. h3 {[%eval -8,11] [%emt 0:00:12]} Rd8 {[%eval -39,35] [%emt 0:00:09]} 23. Rd4 {[%eval -8,10] [%emt 0:00:02]} Ba8 {[%eval -30,33] [%emt 0:00:08] (Qe7)} 24. Kh2 {[%eval -1,9] [%emt 0:00:20] (Dd2)} Qe7 {[%eval -29,29] [%emt 0:00:08]} 25. a4 {[%eval -3,12] [%emt 0:00:05] (De3)} Re5 {[%eval -24,28] [%emt 0:00:07] (g5)} 26. axb5 { [%eval -9,9] [%emt 0:00:10]} axb5 {[%eval -38,31] [%emt 0:00:04]} 27. Rb4 { [%eval -13,10] [%emt 0:00:15] (b4)} Bc6 {[%eval -38,26] [%emt 0:00:03] (g5)} 28. Rbd4 {[%eval -14,9] [%emt 0:00:05] (Dd2)} g5 {[%eval -34,31] [%emt 0:00:14] } 29. Qg3 {[%eval -22,11] [%emt 0:00:00] (Dd2)} h5 {[%eval -55,28] [%emt 0:00: 11] (Kg7)} 30. Kg1 {[%eval -40,10] [%emt 0:00:20] (Df2)} Kg7 {[%eval -119,25] [%emt 0:00:04]} 31. R1d3 {[%eval -52,15] [%emt 0:00:07] (Tf1)} h4 {[%eval -223, 34] [%emt 0:00:23] (Rc8)} 32. Qf2 {[%eval -60,14] [%emt 0:00:05]} b4 {[%eval -203,35] [%emt 0:00:00]} 33. Rxb4 {[%eval -89,17] [%emt 0:00:05]} d5 {[%eval -206,37] [%emt 0:00:00]} 34. exd5 {[%eval -60,20] [%emt 0:00:08]} Qxb4 { [%eval -206,39] [%emt 0:00:00]} 35. dxc6 {[%eval -65,23] [%emt 0:00:03]} Rde8 { [%eval -206,40] [%emt 0:00:02]} 36. Qd4 {[%eval -68,21] [%emt 0:00:02]} Qb8 { [%eval -206,40] [%emt 0:00:02]} 37. Kf2 {[%eval -73,23] [%emt 0:00:07]} Qxb2 { [%eval -206,44] [%emt 0:00:00]} 38. c7 {[%eval -74,22] [%emt 0:00:04]} R5e7 { [%eval -207,42] [%emt 0:00:00]} 39. Nd5 {[%eval -75,21] [%emt 0:00:04]} Qxd4+ { [%eval -235,41] [%emt 0:00:02]} 40. Rxd4 {[%eval -77,20] [%emt 0:00:02]} Nxd5 { [%eval -255,37] [%emt 0:00:02]} 41. Rxd5 {[%eval -78,18] [%emt 0:00:02]} Rxc7 { [%eval -273,37] [%emt 0:00:00]} 42. Rxg5+ {[%eval -79,16] [%emt 0:00:04]} Kf6 { [%eval -283,39] [%emt 0:00:00]} 43. Rg4 {[%eval -79,15] [%emt 0:00:04]} Rxc2+ { [%eval -308,35] [%emt 0:00:04]} 44. Kg1 {[%eval -81,13] [%emt 0:00:17]} Rh8 { [%eval -331,39] [%emt 0:00:00] (Rc1+)} 45. Rf4+ {[%eval -76,10] [%emt 0:00:29]} Ke6 {[%eval -350,47] [%emt 0:00:00]} 46. Re4+ {[%eval -77,11] [%emt 0:00:10] (Ag4+)} Kf5 {[%eval -360,35] [%emt 0:00:03]} 47. Rd4 {[%eval -80,10] [%emt 0: 00:16] (Tb4)} Rb2 {[%eval -393,36] [%emt 0:00:09] (Rc1+)} 48. Ra4 {[%eval -76, 8] [%emt 0:00:11]} f6 {[%eval -403,36] [%emt 0:00:00]} 49. Ra8 {[%eval -75,9] [%emt 0:00:05] (Ta5+)} Rxa8 {[%eval -461,62] [%emt 0:00:03] (Rh6)} 50. Bxa8 { [%eval -54,8] [%emt 0:00:03]} Rb1+ {[%eval -461,70] [%emt 0:00:00] (Kf4)} 51. Kh2 {[%eval -48,8] [%emt 0:00:04]} Rb3 {[%eval -461,74] [%emt 0:00:00] (Ke5)} 52. Bd5 {[%eval -53,8] [%emt 0:00:08] (Ac6)} Re3 {[%eval -461,68] [%emt 0:00: 04] (Rg3)} 53. Bc6 {[%eval -44,7] [%emt 0:00:04] (Ac4)} Rb3 {[%eval -461,75] [%emt 0:00:03] (Kf4)} 54. Bd5 {[%eval -33,7] [%emt 0:00:03] (Aa8)} Re3 { [%eval -461,70] [%emt 0:00:03] (Rg3)} 55. Ba8 {[%eval -45,7] [%emt 0:00:04] (Ac4)} Kg5 {[%eval -461,65] [%emt 0:00:03] (Ke5)} 56. Bb7 {[%eval -43,7] [%emt 0:00:03] (Rg1)} Kf4 {[%eval -461,72] [%emt 0:00:03] (f5)} 57. Ba8 {[%eval -55, 7] [%emt 0:00:02] (Ac6)} Rb3 {[%eval -461,72] [%emt 0:00:03] (f5)} 58. Bc6 { [%eval -58,8] [%emt 0:00:02] (Ad5)} Rg3 {[%eval -461,71] [%emt 0:00:03] (f5)} 59. Ba8 {[%eval -68,7] [%emt 0:00:02] (Ad5)} Ra3 {[%eval -461,71] [%emt 0:00: 04] (f5)} 60. Bb7 {[%eval -56,7] [%emt 0:00:02] (Ad5)} Re3 {[%eval -461,72] [%emt 0:00:03] (f5)} 61. Ba8 {[%eval -54,7] [%emt 0:00:01] (Ac6)} Rb3 {[%eval -461,72] [%emt 0:00:02] (f5)} 62. Bd5 {[%eval -53,8] [%emt 0:00:01]} Rb2 { [%eval -461,67] [%emt 0:00:01] (Rg3)} 63. Bc6 {[%eval -48,7] [%emt 0:00:01] (Ae6)} Rb1 {[%eval -461,66] [%emt 0:00:04] (f5)} 64. Ba8 {[%eval -43,6] [%emt 0:00:01] (Af3)} Re1 {[%eval -461,65] [%emt 0:00:03] (f5)} 65. Bb7 {[%eval -43, 6] [%emt 0:00:01] (Ad5)} Re2 {[%eval -461,68] [%emt 0:00:23] (f5)} 66. Bf3 { [%eval -30,6] [%emt 0:00:01]} Re5 {[%eval -461,65] [%emt 0:00:00] (Rf2)} 67. Bb7 {[%eval -37,6] [%emt 0:00:01] (Ag4)} Rf5 {[%eval -461,63] [%emt 0:00:02] (f5)} 68. Ba8 {[%eval -38,6] [%emt 0:00:01] (Ac6)} Rc5 {[%eval -461,60] [%emt 0:00:17] (Ke3)} 69. Bb7 {[%eval -38,7] [%emt 0:00:00] (Af3)} Kg5 {[%eval -461, 59] [%emt 0:00:13] (Ke3)} 70. Ba8 {[%eval -27,6] [%emt 0:00:01] (Af3)} Kf4 { [%eval -461,55] [%emt 0:00:09] (f5)} 71. Bb7 {[%eval -38,7] [%emt 0:00:00] (Af3)} Ke3 {[%eval -461,56] [%emt 0:00:07] (f5)} 72. Ba8 {[%eval -55,7] [%emt 0:00:00] (Af3)} Rc3 {[%eval -461,53] [%emt 0:00:05] (Rg5)} 73. Bb7 {[%eval -45, 6] [%emt 0:00:00] (Ad5)} Kf2 {[%eval -461,53] [%emt 0:00:04] (f5)} 74. Be4 { [%eval -47,6] [%emt 0:00:00]} Rc1 {[%eval -461,49] [%emt 0:00:01] (Rg3)} 75. Ba8 {[%eval -45,6] [%emt 0:00:00] (Ad3)} Rc3 {[%eval -461,46] [%emt 0:00:01] (f5)} 76. Be4 {[%eval -39,6] [%emt 0:00:00]} Rc1 {[%eval -461,42] [%emt 0:00: 01] (Rg3)} 77. Ba8 {[%eval -37,5] [%emt 0:00:00] (Ad3)} Ke2 {[%eval -461,44] [%emt 0:00:02] (f5)} 78. Be4 {[%eval -23,5] [%emt 0:00:00] (Af3+)} Kf1 { [%eval -461,41] [%emt 0:00:02] (Ke3)} 79. Ba8 {[%eval -26,6] [%emt 0:00:00] (Ag6)} Kf2 {[%eval -461,29] [%emt 0:00:00] (f5)} 80. Be4 {[%eval -43,6] [%emt 0:00:00]} Rc3 {[%eval -461,42] [%emt 0:00:00] (Re1)} 81. Bb7 {[%eval -53,7] [%emt 0:00:00] (Af5)} Rg3 {[%eval -461,36] [%emt 0:00:01] (f5)} 82. Be4 { [%eval -54,6] [%emt 0:00:00]} Ke3 {[%eval -461,34] [%emt 0:00:00] (f5)} 83. Ba8 {[%eval -58,6] [%emt 0:00:00] (Af5)} Rg7 {[%eval -461,30] [%emt 0:00:01] (f5)} 84. Bf3 {[%eval -45,6] [%emt 0:00:00] (Ac6)} Rg3 {[%eval -461,16] [%emt 0:00: 00] (f5)} 85. Ba8 {[%eval -54,6] [%emt 0:00:00] (Ad5)} Kd4 {[%eval -461,22] [%emt 0:00:00] (f5)} 86. Bb7 {[%eval -41,6] [%emt 0:00:00] (Ac6)} Rb3 {[%eval -461,24] [%emt 0:00:00] (f5)} 87. Ba8 {[%eval -29,6] [%emt 0:00:00] (Ac6)} Kc5 {[%eval -461,26] [%emt 0:00:00] (f5)} 88. Be4 {[%eval -22,5] [%emt 0:00:00]} Kd6 {[%eval -453,19] [%emt 0:00:00] (Ra3)} 89. Ba8 {[%eval -15,5] [%emt 0:00: 00] (Rg1)} Ra3 {[%eval -454,14] [%emt 0:00:00] (f5)} 90. Bb7 {[%eval -18,5] [%emt 0:00:00]} Rg3 {[%eval -461,15] [%emt 0:00:00] (f5)} 91. Ba8 {[%eval -10, 5] [%emt 0:00:00] (Rg1)} Re3 {[%eval -443,10] [%emt 0:00:00] (Ra3)} 92. Bb7 { [%eval -19,6] [%emt 0:00:00] (Rg1)} Kd7 {[%eval -461,10] [%emt 0:00:00] (f5)} 93. Ba8 {[%eval -10,5] [%emt 0:00:00]} Rc3 {[%eval -326,29] [%emt 0:00:00] (f5) } 94. Bb7 {[%eval -5,4] [%emt 0:00:00]} Rc4 {[%eval -300,42] [%emt 0:00:00] (f5)} 95. Ba8 {[%eval -6,4] [%emt 0:00:00] (Ad5)} Ke6 {[%eval -300,18] [%emt 0: 00:00] (f5)} 96. Bb7 {[%eval -7,5] [%emt 0:00:00]} Rf4 {[%eval -300,59] [%emt 0:00:00] (f5)} 97. Ba8 {[%eval -5,5] [%emt 0:00:00] (Ac6)} f5 {[%eval -300,19] [%emt 0:00:00]} 98. Bb7 {[%eval -54,7] [%emt 0:00:00] (Ac6)} Rf1 {[%eval -300, 19] [%emt 0:00:00] (Kd6)} 99. Ba8 {[%eval -50,6] [%emt 0:00:00] (Aa6)} Re1 { [%eval -300,19] [%emt 0:00:00] (Kd6)} 100. Bb7 {[%eval -47,6] [%emt 0:00:00] (Af3)} Ke7 {[%eval -300,20] [%emt 0:00:00] (Ke5)} 101. Ba8 {[%eval -41,6] [%emt 0:00:00] (Lag: Av=0.72s, max=1.4s)} 1-0 [/pgn]
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis
shrapnel
Posts: 1339
Joined: Fri Nov 02, 2012 9:43 am
Location: New Delhi, India

Re: Twin 2080Tis and large NN

Post by shrapnel »

Oh, here's another, where my bada$$ 2080 Tis gang up to beat the Brains out of the latest, much-vaunted Brainfish....BRUTAL... :)
[pgn][Event "Rated game, 5 min"] [Site "Engine Room"] [Date "2019.10.29"] [Round "?"] [White "Shrapnel, Lc0 v0.22.0"] [Black "Schinitzu, Brainfish 271019 64"] [Result "1-0"] [ECO "B72"] [WhiteElo "2431"] [BlackElo "2443"] [Annotator "0.29;0.00"] [PlyCount "121"] [EventDate "2019.10.29"] [SourceTitle "playchess.com"] [TimeControl "300"] {Brainfish 271019 64 POPCNT (7 threads): 27.9 plies; 8,272kN/s Intel(R) Core (TM) i7-4720HQ CPU @ 2.60GHz 2593MHz, (4 cores, 8 threads), Samurai.ctg, 32768 MB} 1. e4 {[%eval 0,0] [%emt 0:00:00]} c5 {[%eval 0,0] [%emt 0:00:00]} 2. Nf3 { [%eval 0,0] [%emt 0:00:00]} d6 {[%eval 0,0] [%emt 0:00:00]} 3. d4 {[%eval 0,0] [%emt 0:00:00]} cxd4 {[%eval 0,0] [%emt 0:00:00]} 4. Nxd4 {[%eval 0,0] [%emt 0: 00:00]} Nf6 {[%eval 0,0] [%emt 0:00:00]} 5. Bd3 {[%eval 0,0] [%emt 0:00:00]} g6 {[%eval 0,0] [%emt 0:00:00]} 6. Nc3 {[%eval 0,0] [%emt 0:00:00]} a6 {[%eval 0, 0] [%emt 0:00:00]} 7. Be3 {[%eval 29,9] [%emt 0:00:06]} Bg7 {[%eval 0,0] [%emt 0:00:00] (Ng4)} 8. f3 {[%eval 42,11] [%emt 0:00:03]} O-O {[%eval 0,0] [%emt 0: 00:00] (Nc6)} 9. Qd2 {[%eval 40,13] [%emt 0:00:04]} Nc6 {[%eval 0,0] [%emt 0: 00:00]} 10. O-O-O {[%eval 42,15] [%emt 0:00:05]} d5 {[%eval 0,0] [%emt 0:00:00] } 11. Nxc6 {[%eval 43,16] [%emt 0:00:03]} bxc6 {[%eval 0,0] [%emt 0:00:00]} 12. Bc4 {[%eval 43,16] [%emt 0:00:04]} e6 {[%eval 0,0] [%emt 0:00:00] (Rb8)} 13. Bb3 {[%eval 43,12] [%emt 0:00:10]} Qc7 {[%eval 0,0] [%emt 0:00:00] (a5)} 14. h4 {[%eval 64,14] [%emt 0:00:07]} dxe4 {[%eval 0,0] [%emt 0:00:00]} 15. h5 { [%eval 60,17] [%emt 0:00:05]} exf3 {[%eval 0,29] [%emt 0:00:15]} 16. h6 { [%eval 49,22] [%emt 0:00:16] (hxg6)} Bh8 {[%eval 0,27] [%emt 0:00:06]} 17. gxf3 {[%eval 44,29] [%emt 0:00:00]} Bb7 {[%eval -2,27] [%emt 0:00:05] (c5)} 18. Bc5 {[%eval 65,16] [%emt 0:00:03] (Lf4)} Rfd8 {[%eval 0,30] [%emt 0:00:12]} 19. Bd6 {[%eval 70,21] [%emt 0:00:00]} Qa5 {[%eval 0,32] [%emt 0:00:04]} 20. Qe3 { [%eval 71,24] [%emt 0:00:02]} Qf5 {[%eval 0,33] [%emt 0:00:03] (Rd7)} 21. Qb6 { [%eval 118,17] [%emt 0:00:13]} Ne8 {[%eval 54,37] [%emt 0:00:33]} 22. Bc5 { [%eval 126,22] [%emt 0:00:00]} Rdb8 {[%eval 10,31] [%emt 0:00:13]} 23. Rd7 { [%eval 127,24] [%emt 0:00:00]} Bc8 {[%eval 76,31] [%emt 0:00:07]} 24. Rd8 { [%eval 131,23] [%emt 0:00:00]} Bd7 {[%eval 42,32] [%emt 0:00:06]} 25. Rxb8 { [%eval 137,21] [%emt 0:00:03]} Qf4+ {[%eval 40,32] [%emt 0:00:04]} 26. Kb1 { [%eval 139,21] [%emt 0:00:03] (Le3)} Qxb8 {[%eval 14,27] [%emt 0:00:04]} 27. Na4 {[%eval 135,25] [%emt 0:00:04]} Nf6 {[%eval 31,28] [%emt 0:00:04]} 28. Rd1 {[%eval 135,20] [%emt 0:00:00] (Da5)} Be8 {[%eval 116,29] [%emt 0:00:10]} 29. Bd6 {[%eval 132,24] [%emt 0:00:00]} Qd8 {[%eval 111,30] [%emt 0:00:03] (Qa7)} 30. Qb7 {[%eval 164,22] [%emt 0:00:28]} Qc8 {[%eval 120,37] [%emt 0:00:10] (Nd7)} 31. Qxc8 {[%eval 208,19] [%emt 0:00:19]} Rxc8 {[%eval 194,34] [%emt 0: 00:12]} 32. Nb6 {[%eval 213,22] [%emt 0:00:01]} Rd8 {[%eval 205,30] [%emt 0:00: 04]} 33. c4 {[%eval 215,19] [%emt 0:00:00]} c5 {[%eval 178,32] [%emt 0:00:08] (e5)} 34. Bxc5 {[%eval 222,17] [%emt 0:00:13]} Rb8 {[%eval 208,34] [%emt 0:00: 08] (Rxd1+)} 35. Ba4 {[%eval 305,12] [%emt 0:00:15]} Bxa4 {[%eval 218,32] [%emt 0:00:00]} 36. Nxa4 {[%eval 302,12] [%emt 0:00:05]} Ne8 {[%eval 219,28] [%emt 0:00:03]} 37. b4 {[%eval 319,12] [%emt 0:00:09] (Le3)} Bf6 {[%eval 323, 30] [%emt 0:00:22]} 38. Nb6 {[%eval 310,15] [%emt 0:00:00]} Bg5 {[%eval 326,26] [%emt 0:00:01]} 39. Nd7 {[%eval 317,13] [%emt 0:00:03]} Rb7 {[%eval 336,33] [%emt 0:00:16] (Rc8)} 40. Ne5 {[%eval 370,11] [%emt 0:00:13] (Lb6)} Rc7 { [%eval 391,27] [%emt 0:00:07]} 41. f4 {[%eval 379,13] [%emt 0:00:05]} Bxh6 { [%eval 444,31] [%emt 0:00:14]} 42. Rd8 {[%eval 372,13] [%emt 0:00:00]} Bxf4 { [%eval 501,29] [%emt 0:00:03]} 43. Nd3 {[%eval 369,13] [%emt 0:00:13] (Txe8+)} Bg3 {[%eval 439,24] [%emt 0:00:05]} 44. b5 {[%eval 370,13] [%emt 0:00:16] (Txe8+)} axb5 {[%eval 380,22] [%emt 0:00:03]} 45. cxb5 {[%eval 370,11] [%emt 0: 00:00] (Txe8+)} Kg7 {[%eval 514,27] [%emt 0:00:14] (h5)} 46. b6 {[%eval 361,11] [%emt 0:00:04] (Txe8)} Rxc5 {[%eval 533,25] [%emt 0:00:06]} 47. Nxc5 {[%eval 348,12] [%emt 0:00:02]} Nd6 {[%eval 621,26] [%emt 0:00:02]} 48. Rd7 {[%eval 351,11] [%emt 0:00:06] (a4)} h5 {[%eval 372,23] [%emt 0:00:04]} 49. b7 { [%eval 365,11] [%emt 0:00:03] (Sxe6+)} Nxb7 {[%eval 316,21] [%emt 0:00:01]} 50. Rxb7 {[%eval 366,11] [%emt 0:00:04]} h4 {[%eval 469,26] [%emt 0:00:00]} 51. Ne4 {[%eval 395,10] [%emt 0:00:07]} Be1 {[%eval 604,27] [%emt 0:00:00] (Bf4)} 52. Kc2 {[%eval 447,9] [%emt 0:00:06] (a4)} Ba5 {[%eval 699,26] [%emt 0:00:05] (h3) } 53. Kd3 {[%eval 822,7] [%emt 0:00:07] (Sd6)} h3 {[%eval 370,18] [%emt 0:00: 01]} 54. Nf2 {[%eval 742,7] [%emt 0:00:06] (Tb1)} h2 {[%eval 793,25] [%emt 0: 00:06]} 55. Ke2 {[%eval 607,8] [%emt 0:00:02] (Tb5)} Bc3 {[%eval 441,22] [%emt 0:00:02] (g5)} 56. Nh1 {[%eval 796,6] [%emt 0:00:02] (a4)} g5 {[%eval 409,20] [%emt 0:00:01]} 57. Kf3 {[%eval 717,6] [%emt 0:00:02]} Kg6 {[%eval 571,24] [%emt 0:00:01] (e5)} 58. Kg2 {[%eval 882,6] [%emt 0:00:02]} f5 {[%eval 677,24] [%emt 0:00:01]} 59. Kxh2 {[%eval 874,6] [%emt 0:00:01]} e5 {[%eval 721,25] [%emt 0:00:00]} 60. Ng3 {[%eval 1292,6] [%emt 0:00:01] (Kg2)} f4 {[%eval 722, 22] [%emt 0:00:01]} 61. Ne4 {[%eval 1376,7] [%emt 0:00:01] (Tb6+) Schinitzu, Brainfish 271019 64 abbandona (Lag: Av=0.39s, max=0.7s)} 1-0[/pgn]
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis