Is EMAN the future number one?

[Dr.Wael Deeb]

From one of my current engine tournaments a clear and deserved winner emerged: EMAN.

Have fun: Walter

Hi Walter
I do not own Eman 5.50 and don't trust myself to make a forecast. Still: 60 games without a loss is very impressive.
Regards
Kurt

Indeed it is ....

Dr.D

[Glarean]

So a losing move was sent as an example of Eman being stellar? Well then...

Ok, here is another quick catch from Eman.
I don't think Stockfish 11 will find the solution so quickly as well because Eman is a very good endgame player, even on my rather weak Ryzen7/16T.

Grandelius vs Raznikov (EU-CH U18 -2011)
[d]8/4nk2/1p3p2/1r1p2pp/1P1R1N1P/6P1/3KPP2/8 w - - 0 1
Analysis by Eman 5.60:
1.Sd3 Sf5 2.e3 Sd6 3.Kc2 Se4 4.Kb2 Ke6 5.Ka3 Kd6 6.Sc1 Sxf2 7.Ka4 [...]
Die Stellung ist ausgeglichen: = (0.08 ++) Tiefe: 47/39 00:00:22 490MN, tb=3133756
Weiss hat entscheidenden Vorteil: +- (3.41) Tiefe: 47/74 00:00:50 1116MN, tb=6986652

And now risk a look at the following regrouping of the tower. Eman only needs 6 seconds for this (on my leisurely Ryzen7):

Lounek vs DeOliveira (Corr-Game 2016)
[d]8/4r1pk/2B5/1P3P2/2PrPb1p/6qP/1R1p2P1/3R2QK b - - 0 1
Analysis by Eman 5.60:
1...Ta7 2.Tbb1 Ta3 3.f6 gxf6 4.b6 Txc4 5.b7 Txc6 6.Txd2 Lxd2 7.b8D
Die Stellung ist ausgeglichen: = (-0.10 ++) Tiefe: 41/29 00:00:06 131MN, tb=110611
Schwarz steht klar auf Gewinn: -+ (-5.73 ++) Tiefe: 43/43 00:00:57 1188MN, tb=287865

Think that's impressive.

[Gejsi Marku]

Great. Then request the source code and share it with us. We can then take his improvements, if any, and add them back to Stockfish.

As an ex-user of Eman I can say that those "improvements" are already freely available in other open source Stockfish forks, and that Eman is mostly a compilation of them, some sort of "best of ideas that never got to Stockfish dev" project, and I believe THE MAIN REASON the sources remain private is that the author has built a reputation in the InfinityChess community (a lot of people use his engine there, some have been using it to win money in their online torunaments) and it would get ruined if the sources were revealed and original code (unique to Eman) was hard to find.

I've praised Eman in the past, because it does play better than Stockfish dev (mainly due to smarter time management that is only apparent when playing with books that may leave the engine in a position sooner or later and it can be worthwhile to spend more time on the very first position and gradually play faster but this will not be seen with Stockfish's testing), but its move choices match Stockfish >90% of the time (if I'm generous, feel free to add as many 9s as necessary.)

I welcome an Eman user to prove me wrong, post a position where Eman finds a move that no other SF fork does, this will show the original work that the author put into it. Even Houdini 6 has positions where Stockfish can't find the move while its Tactical setting can, but Eman? Calling it a "quasi-independent engine" sounds like a scam.

I use Eman. And I have definitely come across multiple times of instances where Eman found moves SF couldn't even after a long search. But let's not get mistaken here. That is not because Eman is different than SF. That is because it's experience file. And depending how big your experience file is. You can call Eman more like a giant book that keeps growing.

[Glarean]

I use Eman. And I have definitely come across multiple times of instances where Eman found moves SF couldn't even after a long search. But let's not get mistaken here. That is not because Eman is different than SF. That is because it's experience file. And depending how big your experience file is.

Not with me. I erase the experience file before every puzzle test by Eman. This is the only way a reasonable comparison is possible.

[zullil]

So a losing move was sent as an example of Eman being stellar? Well then...

Ok, here is another quick catch from Eman.
I don't think Stockfish 11 will find the solution so quickly as well because Eman is a very good endgame player, even on my rather weak Ryzen7/16T.

Grandelius vs Raznikov (EU-CH U18 -2011)
[d]8/4nk2/1p3p2/1r1p2pp/1P1R1N1P/6P1/3KPP2/8 w - - 0 1
Analysis by Eman 5.60:
1.Sd3 Sf5 2.e3 Sd6 3.Kc2 Se4 4.Kb2 Ke6 5.Ka3 Kd6 6.Sc1 Sxf2 7.Ka4 [...]
Die Stellung ist ausgeglichen: = (0.08 ++) Tiefe: 47/39 00:00:22 490MN, tb=3133756
Weiss hat entscheidenden Vorteil: +- (3.41) Tiefe: 47/74 00:00:50 1116MN, tb=6986652

Stockfish-dev solves this quickly:

info depth 47 seldepth 41 multipv 1 score cp 10 lowerbound nodes 586626897 nps 29477257 hashfull 252 tbhits 5019909 time 19901 pv f4d3
info depth 46 currmove f4d3 currmovenumber 1
info depth 47 seldepth 46 multipv 1 score cp 20 lowerbound nodes 602474102 nps 29607061 hashfull 257 tbhits 5107895 time 20349 pv f4d3
info depth 45 currmove f4d3 currmovenumber 1
info depth 47 seldepth 46 multipv 1 score cp 35 lowerbound nodes 612178943 nps 29708771 hashfull 260 tbhits 5143169 time 20606 pv f4d3
info depth 44 currmove f4d3 currmovenumber 1
info depth 47 seldepth 52 multipv 1 score cp 56 lowerbound nodes 629514636 nps 29815034 hashfull 266 tbhits 5273161 time 21114 pv f4d3
info depth 43 currmove f4d3 currmovenumber 1
info depth 47 seldepth 52 multipv 1 score cp 84 lowerbound nodes 643258878 nps 29860685 hashfull 271 tbhits 5416388 time 21542 pv f4d3
info depth 42 currmove f4d3 currmovenumber 1
info depth 47 seldepth 52 multipv 1 score cp 121 lowerbound nodes 654714289 nps 29928427 hashfull 275 tbhits 5485090 time 21876 pv f4d3
info depth 41 currmove f4d3 currmovenumber 1
info depth 47 seldepth 52 multipv 1 score cp 170 lowerbound nodes 665680760 nps 30018071 hashfull 279 tbhits 5553422 time 22176 pv f4d3
info depth 40 currmove f4d3 currmovenumber 1
info depth 47 seldepth 56 multipv 1 score cp 234 lowerbound nodes 708140252 nps 30340199 hashfull 294 tbhits 5879063 time 23340 pv f4d3
info depth 39 currmove f4d3 currmovenumber 1
info depth 47 seldepth 56 multipv 1 score cp 316 lowerbound nodes 738935043 nps 30595190 hashfull 306 tbhits 6074680 time 24152 pv f4d3
info depth 38 currmove f4d3 currmovenumber 1
info depth 47 seldepth 70 multipv 1 score cp 420 lowerbound nodes 1011547375 nps 31807665 hashfull 387 tbhits 8162542 time 31802 pv f4d3
info depth 37 currmove f4d3 currmovenumber 1
info depth 37 currmove f4h5 currmovenumber 2
info depth 37 currmove h4g5 currmovenumber 3
info depth 37 currmove f4g2 currmovenumber 4
info depth 37 currmove e2e3 currmovenumber 5
info depth 37 currmove d2e1 currmovenumber 6
info depth 37 currmove d2c2 currmovenumber 7
info depth 37 currmove f2f3 currmovenumber 8
info depth 37 currmove d2c3 currmovenumber 9
info depth 37 currmove d2d3 currmovenumber 10
info depth 37 currmove d4d3 currmovenumber 11
info depth 37 currmove g3g4 currmovenumber 12
info depth 37 currmove f4e6 currmovenumber 13
info depth 37 currmove d4c4 currmovenumber 14
info depth 37 currmove d4e4 currmovenumber 15
info depth 37 currmove f4g6 currmovenumber 16
info depth 37 currmove d2c1 currmovenumber 17
info depth 37 currmove d2e3 currmovenumber 18
info depth 37 currmove e2e4 currmovenumber 19
info depth 37 currmove d2d1 currmovenumber 20
info depth 37 currmove f4h3 currmovenumber 21
info depth 37 currmove f4d5 currmovenumber 22
info depth 37 currmove d4d5 currmovenumber 23
info depth 47 seldepth 70 multipv 1 score cp 536 nodes 1917090274 nps 34668347 hashfull 599 tbhits 13497866 time 55298 pv f4d3 e7f5 e2e3 f5d6 d2c2 d6e4 d3c1 e4c5 c1e2 c5a6 c2b3 a6b4 d4b4 b5a5 b3b2 a5a6 e2c3 a6a8 b4b6 a8d8 c3e2 d8c8 b2b3 f7g6 e2c3 c8d8 h4g5 g6g5 b6b4 g5g6 b4d4 d8b8 b3c2 b8a8 d4d5 a8a1 c3e2 a1f1 e2f4 g6h6 d5h5 h6g7

[zullil]

So a losing move was sent as an example of Eman being stellar? Well then...

Ok, here is another quick catch from Eman.
I don't think Stockfish 11 will find the solution so quickly as well because Eman is a very good endgame player, even on my rather weak Ryzen7/16T.

And now risk a look at the following regrouping of the tower. Eman only needs 6 seconds for this (on my leisurely Ryzen7):

Lounek vs DeOliveira (Corr-Game 2016)
[d]8/4r1pk/2B5/1P3P2/2PrPb1p/6qP/1R1p2P1/3R2QK b - - 0 1
Analysis by Eman 5.60:
1...Ta7 2.Tbb1 Ta3 3.f6 gxf6 4.b6 Txc4 5.b7 Txc6 6.Txd2 Lxd2 7.b8D
Die Stellung ist ausgeglichen: = (-0.10 ++) Tiefe: 41/29 00:00:06 131MN, tb=110611
Schwarz steht klar auf Gewinn: -+ (-5.73 ++) Tiefe: 43/43 00:00:57 1188MN, tb=287865

Think that's impressive.

Stockfish-dev:

info depth 44 seldepth 35 multipv 1 score cp 10 lowerbound nodes 353471429 nps 34990242 hashfull 146 tbhits 678578 time 10102 pv e7a7
info depth 43 currmove e7a7 currmovenumber 1
info depth 44 seldepth 35 multipv 1 score cp 20 lowerbound nodes 356568168 nps 34988535 hashfull 147 tbhits 681220 time 10191 pv e7a7
info depth 42 currmove e7a7 currmovenumber 1
info depth 44 seldepth 35 multipv 1 score cp 35 lowerbound nodes 369610486 nps 34984428 hashfull 151 tbhits 695717 time 10565 pv e7a7
info depth 41 currmove e7a7 currmovenumber 1
info depth 44 seldepth 35 multipv 1 score cp 56 lowerbound nodes 372455962 nps 34978959 hashfull 152 tbhits 699503 time 10648 pv e7a7
info depth 40 currmove e7a7 currmovenumber 1
info depth 44 seldepth 35 multipv 1 score cp 84 lowerbound nodes 384257464 nps 34961101 hashfull 159 tbhits 711640 time 10991 pv e7a7
info depth 39 currmove e7a7 currmovenumber 1
info depth 44 seldepth 58 multipv 1 score cp 122 lowerbound nodes 445117193 nps 35070689 hashfull 183 tbhits 756741 time 12692 pv e7a7
info depth 38 currmove e7a7 currmovenumber 1
info depth 44 seldepth 59 multipv 1 score cp 171 lowerbound nodes 518611429 nps 35229361 hashfull 209 tbhits 817999 time 14721 pv e7a7
info depth 37 currmove e7a7 currmovenumber 1
info depth 44 seldepth 59 multipv 1 score cp 234 lowerbound nodes 559787657 nps 35173588 hashfull 227 tbhits 848407 time 15915 pv e7a7
info depth 36 currmove e7a7 currmovenumber 1
info depth 44 seldepth 59 multipv 1 score cp 316 lowerbound nodes 647706736 nps 35207193 hashfull 259 tbhits 906215 time 18397 pv e7a7
info depth 35 currmove e7a7 currmovenumber 1
info depth 44 seldepth 59 multipv 1 score cp 420 lowerbound nodes 755466002 nps 35277422 hashfull 304 tbhits 958649 time 21415 pv e7a7
info depth 34 currmove e7a7 currmovenumber 1
info depth 34 currmove d4c4 currmovenumber 2
info depth 34 currmove f4e3 currmovenumber 3
info depth 34 currmove f4e5 currmovenumber 4
info depth 34 currmove g3h2 currmovenumber 5
info depth 34 currmove h7h6 currmovenumber 6
info depth 34 currmove h7h8 currmovenumber 7
info depth 34 currmove g3e1 currmovenumber 8
info depth 34 currmove g3e3 currmovenumber 9
info depth 34 currmove h7g8 currmovenumber 10
info depth 34 currmove g7g5 currmovenumber 11
info depth 34 currmove g7g6 currmovenumber 12
info depth 34 currmove d4d8 currmovenumber 13
info depth 34 currmove d4d3 currmovenumber 14
info depth 34 currmove g3c3 currmovenumber 15
info depth 34 currmove g3a3 currmovenumber 16
info depth 34 currmove e7e6 currmovenumber 17
info depth 34 currmove f4d6 currmovenumber 18
info depth 34 currmove d4d5 currmovenumber 19
info depth 34 currmove d4d7 currmovenumber 20
info depth 34 currmove g3b3 currmovenumber 21
info depth 34 currmove f4b8 currmovenumber 22
info depth 34 currmove e7e5 currmovenumber 23
info depth 34 currmove e7e8 currmovenumber 24
info depth 34 currmove f4h6 currmovenumber 25
info depth 34 currmove g3g6 currmovenumber 26
info depth 34 currmove g3g5 currmovenumber 27
info depth 34 currmove g3d3 currmovenumber 28
info depth 34 currmove d4d6 currmovenumber 29
info depth 34 currmove g3f2 currmovenumber 30
info depth 34 currmove f4c7 currmovenumber 31
info depth 34 currmove e7c7 currmovenumber 32
info depth 34 currmove e7d7 currmovenumber 33
info depth 34 currmove g3f3 currmovenumber 34
info depth 34 currmove g3g4 currmovenumber 35
info depth 34 currmove e7b7 currmovenumber 36
info depth 34 currmove e7f7 currmovenumber 37
info depth 34 currmove f4g5 currmovenumber 38
info depth 34 currmove g3g2 currmovenumber 39
info depth 34 currmove g3h3 currmovenumber 40
info depth 34 currmove e7e4 currmovenumber 41
info depth 34 currmove d4e4 currmovenumber 42
info depth 44 seldepth 77 multipv 1 score cp 516 nodes 968594582 nps 35516081 hashfull 374 tbhits 1100842 time 27272 pv e7a7 b2b1 a7a3 c6d5 a3e3 e4e5 e3e5 d1f1 d4d3 b5b6 f4e3 g1h2 e3b6 h2g3 h4g3 b1d1 b6f2 d5f3 e5e1 f1e1 d2e1q d1e1 f2e1 h1g1 e1f2 g1f1 d3e3 f3e2 h7h6 c4c5 e3c3 e2a6 c3c5 f1e2 c5f5 a6d3 f5e5 e2d1 h6g5 d1c2 g5f4 c2c3

[Glarean]

Stockfish-dev solves this quickly:
info depth 47 seldepth 70 multipv 1 score cp 536 nodes 1917090274 nps 34668347 hashfull 599 tbhits 13497866 time 55298 pv f4d3 e7f5 e2e3 f5d6 [...]

You forgot some details:
- Which machine?
- Which Stockfish-"Derivative"?
- Which means: "time 55298"?

[Alayan]

Stockfish-dev means the latest development version of Stockfish.

The uci time info is in milliseconds, so the move was found at 19s and the fail-high completed at 55s.

I'll let Zulli answer on the hardware he used.

[zullil]

Stockfish-dev solves this quickly:
info depth 47 seldepth 70 multipv 1 score cp 536 nodes 1917090274 nps 34668347 hashfull 599 tbhits 13497866 time 55298 pv f4d3 e7f5 e2e3 f5d6 [...]

You forgot some details:
- Which machine?
- Which Stockfish-"Derivative"?
- Which means: "time 55298"?

I omitted details to prompt you to respond.

20 threads at 3.15 GHz with 16 GB hash and 6-man tables. Times are in milliseconds.

Of course, because multi-threaded searches are quite non-deterministic, times for individual searches are quite variable. Whether or not EMAN is faster than Stockfish (and I doubt it is any significant way), what I hope has now been made clear to you is that EMAN is a Stockfish-derivative that is being distributed by its "creator" in what seems to be clear violation of Stockfish's GPL.

Have you requested the source code to which you are legally entitled? If so, have you received it?

[Glarean]

I omitted details to prompt you to respond.

Of course...

20 threads at 3.15 GHz with 16 GB hash and 6-man tables. Times are in milliseconds.

Ohlala: 20 Threads & 6-men TB?
My tiny Ryzen7 had only 16 Thr. & only 5-men TB. So my Eman with his 22 seconds is a little faster than your SF...

Whether or not EMAN is faster than Stockfish (and I doubt it is any significant way), what I hope has now been made clear to you is that EMAN is a Stockfish-derivative

What's your question?! The programmer Khalid Omar himself says that Eman is a stockfish derivative: https://eman.zohosites.com/

Have you requested the source code to which you are legally entitled? If so, have you received it?

Are you sure he has to make the source code available to people who didn't get a binary from him? And if people got the binary from somewhere (not from him) anyway, he's probably not obligated to give them the source code, is he?
But I'm not a GPL expert, and I'm not the lawyer of the Eman programmer. The topic bores me a bit, and I'm out of the discussion.

Have fun: Walter