Ethereal - question

Krzysztof Grzelak · Post by **Krzysztof Grzelak** » Sat Jan 01, 2022 2:02 pm

A question for the author of the chess engine Ethereal. How the program is doing while playing when no opening book is used.

Chessqueen · Post by **Chessqueen** » Sat Jan 01, 2022 5:06 pm

Krzysztof Grzelak wrote: ↑Sat Jan 01, 2022 2:02 pm A question for the author of the chess engine Ethereal. How the program is doing while playing when no opening book is used.

Also when there is no EGTB out of Stockfish Komodo Dragon 2.6 Ethereal took the longest to force mate with this position, after Black side resigned it say that there is mate in 14, Stockfish and Komodo Dragon 2.6 solved it with the same amount of moves without using EGTB

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2022.01.01"]
[Round "?"]
[White "Ethereal-pext-avx2"]
[Black "Ethereal-pext-avx2"]
[Result "*"]
[BlackElo "3300"]
[Time "09:53:22"]
[WhiteElo "3300"]
[TimeControl "180+3"]
[SetUp "1"]
[FEN "8/3N2k1/8/8/2B4K/8/8/8 b - - 0 1"]
[Termination "normal"]
[PlyCount "25"]
[WhiteType "program"]
[BlackType "program"]

1. ... Kh6 2. Bf7 Kg7 3. Ba2 Kg6 4. Kg4 Kh7 5. Kg5 Kg7 6. Bb3 Kh8 7. Ne5
Kg7 8. Ng6 Kh7 9. Kf6 Kh6 10. Bg8 Kh5 11. Ne5 Kh4 12. Bd5 Kh5 13. Be4 Kh6
{Black resigns} *[/pgn]

Krzysztof Grzelak · Post by **Krzysztof Grzelak** » Sat Jan 01, 2022 6:08 pm

I played a match with the Revenge 2.0 program. The Ethereal 13.25 program does not play very well without the opening book - it cannot play.
And I interrupted the match.

Parties of the match Revenge 2.0 vs Ethereal 13.25 -https://files.fm/u/j48ynbcas

Guenther · Post by **Guenther** » Sat Jan 01, 2022 6:29 pm

Krzysztof Grzelak wrote: ↑Sat Jan 01, 2022 6:08 pm I played a match with the Revenge 2.0 program. The Ethereal 13.25 program does not play very well without the opening book - it cannot play.
And I interrupted the match.

Parties of the match Revenge 2.0 vs Ethereal 13.25 -https://files.fm/u/j48ynbcas

IIRC you never answered, if you use the HCE or the commercial NNUE version of Ethereal?
If it's the free HCE version the result looks not too dramatic vs. new Revenge.

OTH always playing all games from the same start position w/o any variation is not the best to do for a test.

Code: Select all

C:\Downloads\Test.pgn/18: 1. e4 {+0.19/25 8.57} e5 {-0.23/27 18.20}
C:\Downloads\Test.pgn/92: 1. d4 {+0.25/26 15.71} Nf6 {-0.20/25 12.98}
C:\Downloads\Test.pgn/172: 1. d4 {+0.19/25 8.26} Nf6 {-0.20/24 10.66}
C:\Downloads\Test.pgn/272: 1. e4 {+0.33/26 19.93} e5 {-0.19/26 10.44}
C:\Downloads\Test.pgn/357: 1. e4 {+0.19/26 8.15} c5 {-0.23/24 9.18}
C:\Downloads\Test.pgn/427: 1. d4 {+0.29/26 24.68} Nf6 {-0.19/25 10.05}
C:\Downloads\Test.pgn/474: 1. e4 {+0.25/26 9.75} c5 {-0.17/26 13.68}
C:\Downloads\Test.pgn/515: 1. e4 {+0.34/25 13.69} e5 {-0.14/27 12.41}
C:\Downloads\Test.pgn/617: 1. e4 {+0.23/26 8.31} c5 {-0.17/27 16.75}
C:\Downloads\Test.pgn/689: 1. d4 {+0.31/27 27.23} Nf6 {-0.19/25 9.76}
C:\Downloads\Test.pgn/748: 1. e4 {+0.26/26 10.61} c5 {-0.18/25 9.69}
C:\Downloads\Test.pgn/818: 1. c4 {+0.30/25 21.55} e5 {-0.10/25 7.35}
C:\Downloads\Test.pgn/880: 1. e4 {+0.20/25 7.26} c5 {-0.12/24 9.16}
C:\Downloads\Test.pgn/963: 1. c4 {+0.24/25 17.43} e5 {-0.19/25 7.05}
C:\Downloads\Test.pgn/1021: 1. d4 {+0.17/25 9.06} Nf6 {-0.14/24 6.85}
C:\Downloads\Test.pgn/1112: 1. d4 {+0.26/25 19.87} Nf6 {-0.19/26 17.66}
C:\Downloads\Test.pgn/1165: 1. d4 {+0.27/25 13.54} Nf6 {-0.17/24 10.07}
C:\Downloads\Test.pgn/1238: 1. c4 {+0.28/25 15.80} e5 {-0.17/25 6.95}
C:\Downloads\Test.pgn/1307: 1. e4 {+0.23/26 9.07} c5 {-0.17/26 14.90}
C:\Downloads\Test.pgn/1405: 1. e4 {+0.35/26 22.32} e6 {-0.20/27 7.99}
C:\Downloads\Test.pgn/1487: 1. e4 {+0.26/24 6.83} c5 {-0.13/25 10.41}
C:\Downloads\Test.pgn/1632: 1. d4 {+0.34/25 15.19} Nf6 {-0.19/25 6.71}
C:\Downloads\Test.pgn/1692: 1. Nf3 {+0.19/25 8.46} Nf6 {-0.22/25 9.56}
C:\Downloads\Test.pgn/1755: 1. e4 {+0.32/24 8.32} c5 {-0.19/26 8.52}
C:\Downloads\Test.pgn/1853: 1. e4 {+0.20/25 7.57} c5 {-0.13/25 10.47}
C:\Downloads\Test.pgn/1907: 1. e4 {+0.27/25 9.36} e5 {-0.19/26 8.46}
C:\Downloads\Test.pgn/1965: 1. e4 {+0.26/26 10.94} c5 {-0.27/28 35.81}
C:\Downloads\Test.pgn/2041: 1. e4 {+0.26/27 23.95} e5 {-0.22/25 8.82}
C:\Downloads\Test.pgn/2114: 1. e4 {+0.26/25 8.22} e6 {-0.14/25 20.63}
C:\Downloads\Test.pgn/2178: 1. e4 {+0.34/26 17.86} e5 {-0.20/26 8.32}
C:\Downloads\Test.pgn/2264: 1. e4 {+0.19/26 7.04} c5 {-0.21/27 18.22}
C:\Downloads\Test.pgn/2354: 1. e4 {+0.26/27 19.07} e5 {-0.19/26 9.02}
C:\Downloads\Test.pgn/2431: 1. e4 {+0.19/25 7.84} e5 {-0.24/28 24.98}
C:\Downloads\Test.pgn/2511: 1. e4 {+0.27/26 17.63} e5 {-0.19/26 7.79}
C:\Downloads\Test.pgn/2571: 1. e4 {+0.20/25 7.06} c5 {-0.18/25 12.39}
C:\Downloads\Test.pgn/2663: 1. c4 {+0.30/25 19.46} e5 {-0.19/26 8.27}
C:\Downloads\Test.pgn/2751: 1. e4 {+0.23/27 11.18} e5 {-0.20/27 19.64}
C:\Downloads\Test.pgn/2823: 1. e4 {+0.22/27 21.84} e6 {-0.20/26 7.48}
C:\Downloads\Test.pgn/2935: 1. e4 {+0.20/25 8.04} c5 {-0.12/25 13.78}
C:\Downloads\Test.pgn/2992: 1. c4 {+0.28/25 23.49} e5 {-0.19/27 13.15}
C:\Downloads\Test.pgn/3059: 1. e4 {+0.26/25 9.58} c5 {-0.13/26 15.52}
C:\Downloads\Test.pgn/3143: 1. e4 {+0.30/27 27.89} e5 {-0.19/26 7.83}
C:\Downloads\Test.pgn/3220: 1. e4 {+0.23/26 10.13} c5 {-0.13/25 12.91}
C:\Downloads\Test.pgn/3272: 1. e4 {+0.32/25 11.33} e6 {-0.19/26 11.13}
C:\Downloads\Test.pgn/3378: 1. e4 {+0.20/25 6.69} c5 {-0.16/25 12.85}
C:\Downloads\Test.pgn/3525: 1. c4 {+0.23/27 27.85} e5 {-0.19/26 8.01}
C:\Downloads\Test.pgn/3591: 1. e4 {+0.19/25 6.85} c5 {-0.22/25 13.32}
C:\Downloads\Test.pgn/3689: 1. d4 {+0.27/26 18.25} d5 {-0.19/26 8.95}
C:\Downloads\Test.pgn/3759: 1. e4 {+0.20/25 8.20} c5 {-0.18/24 8.96}
C:\Downloads\Test.pgn/3919: 1. e4 {+0.30/26 22.20} e5 {-0.19/26 8.63}
C:\Downloads\Test.pgn/4058: 1. d4 {+0.19/26 9.17} Nf6 {-0.11/26 25.97}
C:\Downloads\Test.pgn/4123: 1. d4 {+0.26/26 18.05} Nf6 {-0.16/25 8.50}
C:\Downloads\Test.pgn/4183: 1. e4 {+0.22/25 7.90} c5 {-0.16/27 23.20}
C:\Downloads\Test.pgn/4248: 1. e4 {+0.26/25 15.30} e5 {-0.16/26 9.58}
C:\Downloads\Test.pgn/4315: 1. e4 {+0.26/26 7.11} e5 {-0.17/27 24.35}
C:\Downloads\Test.pgn/4370: 1. c4 {+0.28/26 20.46} e5 {-0.16/25 7.61}
C:\Downloads\Test.pgn/4433: 1. d4 {+0.19/25 7.63} Nf6 {-0.23/25 10.90}

From your pgn (the last unfinished game deleted)

Code: Select all

My Tournament
?, 2021.12.30
                          Score     12345678901234567890123456789012345678901234567890       
----------------------------------------------------------------------------------------------
 1: Revenge 2.0 x64     34.0 / 57   =======0====1=1===1=1=======1=1=1=11====1===1===1========  (+12 -1 =44)
 2: Ethereal 13.25 x64  23.0 / 57   =======1====0=0===0=0=======0=0=0=00====0===0===0========  (+1 -12 =44)
----------------------------------------------------------------------------------------------

Krzysztof Grzelak · Post by **Krzysztof Grzelak** » Sat Jan 01, 2022 6:46 pm

Hi Guenther,
both programs are commercial. Ethereal uses commercial NNUE version. Write down why the test without the book is bad. Only then can the strength of the program be known.

Guenther · Post by **Guenther** » Sat Jan 01, 2022 6:56 pm

Krzysztof Grzelak wrote: ↑Sat Jan 01, 2022 6:46 pm Hi Guenther,
both programs are commercial. Ethereal uses commercial NNUE version.

Then it's most likely, because the sample size is much too small (and you are always playing from the same position).
Why not using a small start pgn/book instead?

Also Revenge 2.0 8cpu currently is tested higher than Ethereal 13.25 8cpu @CCRL, but also with extreme error bars though.

Krzysztof Grzelak · Post by **Krzysztof Grzelak** » Sat Jan 01, 2022 7:11 pm

So I wonder if the Ethereal program is well written by the programmer. Because I have the impression that the programmer has never tested his engine without a book.

Luecx · Post by **Luecx** » Sat Jan 01, 2022 9:09 pm

Krzysztof Grzelak wrote: ↑Sat Jan 01, 2022 7:11 pm So I wonder if the Ethereal program is well written by the programmer. Because I have the impression that the programmer has never tested his engine without a book.

xD

Guenther · Post by **Guenther** » Sat Jan 01, 2022 10:33 pm

Krzysztof Grzelak wrote: ↑Sat Jan 01, 2022 7:11 pm So I wonder if the Ethereal program is well written by the programmer. Because I have the impression that the programmer has never tested his engine without a book.

A few things:
If you test from the same position always, you completely defeat the purpose of any statistical meaningful rating calculation.
The games are not independent, this just produces statistical noise then. Actually it is the same like playing just a few games instead.
This was mentioned dozens of times surely over the last 20 years in this forum, just do some research, I won't explain it here
again in detail.

But, I also noticed that something is very suspicious in your games. I took one example with a strange eval (depth also surprisingly low BTW)
and despite not having the commercial NNUE version here, my compilation of 13.43 HCE (not much different from 13.25 HCE wise) has nearly
the same evaluation than Ethereal NNUE in your game??? (on a slow hardware)
As long as I don't see any debug, which confirms you really use the nnue file, I would believe you didn't use it at all in your games...

Ethereal 13.43 SSSE3 GS
1kqr3r/ppp5/1b3p2/6p1/P2P4/2P3P1/3BPPQ1/1RR2K2 w - - 5 28

Code: Select all

dep	score	nodes	time	(not shown:  tbhits	knps	seldep)
 27	 -0.01 	70.1M  	1:35.70	b1a1 b6a5 a1b1 c7c6 e2e4 c8e6 f1g1 h8h7 d4d5 c6d5 e4d5 d8d5 d2e3 d5e5 e3d4 e5e4 d4f6 g5g4 f6d4 e4e2 b1b5 a5b6 c1b1 e2e1 b1e1 e6e1 g2f1 e1e4 f1g2 e4e2 d4b6 a7b6 b5b1 
 26	+1.06 	22.0M  	0:28.95	e2e3 h8h3 f1e2 c7c6 c3c4 c8e6 d4d5 c6d5 a4a5 b6c7 b1b5 e6g4 g2f3 g4f3 e2f3 h3h2 a5a6 b7b6 b5d5 g5g4 f3g4 h2f2 d5d8 c7d8 d2c3 f6f5 g4h3 f2f3 c3d4 d8c7 c1g1 c7d6 h3h4 b8c8 g3g4 f5g4 h4g4 f3f8 g1d1 d6c7 g4g5 f8g8 g5f6 
 26	+1.12 	18.4M  	0:24.08	e2e3 
 26	+1.17 	16.8M  	0:21.99	e2e3 
 25	+1.07 	15.9M  	0:20.88	e2e3 h8h3 f1e2 c7c6 c3c4 c8e6 d4d5 c6d5 a4a5 b6c7 a5a6 b7b6 b1b4 h3h7 d2c3 f6f5 c4d5 e6d5 g2d5 d8d5 b4c4 f5f4 g3f4 g5f4 e3e4 d5b5 c3f6 h7d7 f6d4 
 25	+1.21 	15.5M  	0:20.32	e2e3 
 25	+0.99 	14.9M  	0:19.54	e2e3 
 25	+1.04 	13.3M  	0:17.42	e2e3 
 24	+0.94 	12.6M  	0:16.42	e2e3 h8h3 f1e2 c7c6 c3c4 c8g4 g2f3 g4e6 e2d3 f6f5 a4a5 b6c7 a5a6 b7b6 c4c5 g5g4 f3g2 h3h6 c5b6 a7b6 b1b4 b8a7 d3e2 e6e4 g2e4 f5e4 b4a4 h6f6 a4b4 a7a6 
 24	+1.07 	12.3M  	0:16.00	e2e3 
 24	+1.11 	11.4M  	0:14.69	e2e3 
 24	+1.16 	10.5M  	0:13.61	e2e3 
 23	+1.06 	9.71M  	0:12.46	e2e3 h8h3 f1e2 c7c6 c3c4 c8g4 g2f3 g4e6 e2d3 f6f5 a4a5 g5g4 f3g2 b6c5 c1h1 c5f8 h1h3 g4h3 g2f3 f8g7 b1h1 d8h8 a5a6 b7b6 d3c3 e6e4 f3e4 f5e4 c3c2 
 23	+1.23 	9.43M  	0:12.10	e2e3 
 23	+1.88 	5.01M  	0:06.19	e2e3 
 22	+1.78 	4.22M  	0:05.21	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b1 b8c8 f1g2 g5g4 c1h1 f6f5 h1h4 h8h4 g3h4 d8h8 b1h1 h8h5 d4c5 b6c5 g2g3 c8d7 g3f4 d7e6 a4a5 
 22	+1.84 	4.11M  	0:05.07	e2e3 
 21	+1.94 	2.74M  	0:03.33	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b1 b8c8 f1g2 g5g4 c1h1 b6c7 b1e1 f6f5 e1b1 c7b6 h1h4 h8h4 g3h4 d8h8 b1h1 h8h7 d4c5 b6c5 g2g3 h7b7 
 21	+1.93 	2.45M  	0:02.99	e2e3 
 20	+1.83 	2.16M  	0:02.63	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b1 b8c8 f1g2 g5g4 c1h1 b6c7 b1e1 f6f5 e1b1 c5c4 h1h8 d8h8 b1b4 c8d7 b4b7 
 20	+1.89 	2.11M  	0:02.57	e2e3 
 19	+1.84 	1.43M  	0:01.73	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b3 b8a8 f1g2 g5g4 b3b1 b6c7 e3e4 c5d4 c3d4 d8d4 c1c6 d4d2 c6c7 
 18	+1.68 	854223	0:01.04	e2e3 c8f5 d2e1 d8e8 a4a5 f5a5 c3c4 a5f5 c4c5 e8e3 c5b6 a7b6 f1g1 e3f3 e1d2 f5g4 b1a1 g4f5 d2e3 c7c6 c1b1 
 17	+1.24 	682086	0:00.84	e2e3 c8f5 d2e1 d8e8 a4a5 f5a5 c3c4 a5f5 c4c5 e8e3 c5b6 a7b6 f1g1 e3a3 b1a1 a3d3 a1a4 d3f3 e1d2 f5d3 
 16	+1.36 	410613	0:00.51	e2e3 c8f5 d2e1 h8e8 f1g1 c7c6 g3g4 f5e6 c3c4 b6c7 b1b2 b8a8 c1b1 c7b6 a4a5 
 15	+1.33 	243352	0:00.31	e2e3 c7c5 b1b5 c8e6 f1e1 e6c4 c1b1 c5d4 c3d4 c4a4 g2f3 b8a8 b5f5 a4d7 f5f6 
 14	+1.01 	201245	0:00.26	e2e3 c7c5 b1b5 c8e6 f1e1 e6c4 c1b1 c5d4 c3d4 c4a4 g2f3 d8f8 
 13	+1.11 	70507  	0:00.09	e2e3 c7c5 b1b5 c8e6 f1e1 e6d7 c1b1 d7c6 g2c6 b7c6 b5b6 a7b6 b1b6 b8a7 b6c6 
 12	+0.98 	45843  	0:00.06	e2e3 c7c5 b1b5 c8e6 c1b1 e6c4 f1e1 c5d4 c3d4 c4a4 b5b6 a7b6 b1b6 
 11	+0.69 	32285  	0:00.04	e2e3 c7c5 b1b5 c8f5 c1b1 f5d3 f1e1 d3f5 
 10	+0.26 	29070  	0:00.04	e2e3 c7c5 b1b5 c8f5 c1b1 f5d3 f1e1 c5d4 c3d4 
  9	+0.44 	23260  	0:00.03	e2e3 c7c6 c3c4 c8e6 f1e2 d8e8 c4c5 
  8	+0.45 	12180  	0:00.01	c1d1 c8f5 b1b5 f5e6

[pgn]
[Event "My Tournament"]
[Site "?"]
[Date "2021.12.30"]
[Round "36"]
[White "Ethereal 13.25 x64"]
[Black "Revenge 2.0 x64"]
[Result "0-1"]
[ECO "A20"]
[GameDuration "00:16:44"]
[GameEndTime "2021-12-30T18:38:05.686 Œrodkowoeuropejski czas stand."]
[GameStartTime "2021-12-30T18:21:21.173 Œrodkowoeuropejski czas stand."]
[Opening "English Opening"]
[PlyCount "135"]
[Termination "adjudication"]
[TimeControl "300+3"]

1. c4 {+0.30/25 19.46} e5 {-0.19/26 8.27}
2. g3 {+0.35/25 8.27} Nf6 {-0.19/27 9.56}
3. Bg2 {+0.29/25 13.65} d5 {-0.19/29 9.06}
4. cxd5 {+0.46/26 9.25} Nxd5 {-0.19/30 8.94}
5. Nf3 {+0.38/26 7.63} Nc6 {-0.19/30 29.27}
6. O-O {+0.47/27 8.76} e4 {-0.17/28 11.85}
7. Ne1 {+0.92/28 16.64} h5 {-0.29/27 9.92}
8. Qa4 {+0.89/27 31.23} Be6 {0.00/27 6.47}
9. Qxe4 {+0.87/29 30.36} Qe7 {0.00/27 10.81}
10. d3 {+0.78/30 44.94} h4 {0.00/27 6.14}
11. Nf3 {+0.42/29 14.75} hxg3 {0.00/28 6.00}
12. hxg3 {+0.45/30 20.89} O-O-O {0.00/31 23.18}
13. Nc3 {+0.41/27 6.61} f6 {0.00/32 9.45}
14. Bd2 {+0.01/26 10.30} Kb8 {0.00/29 6.44}
15. Rfc1 {+0.56/27 6.11} Qd7 {-0.16/30 23.65}
16. Nd4 {+0.66/27 5.45} Nxd4 {0.00/35 5.22}
17. Qxd4 {+0.73/26 3.86} Nxc3 {0.00/35 6.47}
18. bxc3 {+0.48/28 7.41} Qc8 {-0.07/34 21.74}
19. Qa4 {+0.40/27 8.91} Bc5 {0.00/29 5.21}
20. Rab1 {+0.01/28 13.51} Bd7 {0.00/32 6.30}
21. Qf4 {+0.18/31 12.94} g5 {0.00/34 6.30}
22. Qf3 {+0.16/30 9.46} Bb6 {0.00/35 5.01}
23. d4 {+0.01/30 9.44} Bh3 {0.00/33 6.13}
24. Bxh3 {+0.32/29 5.79} Qxh3 {0.00/32 8.02}
25. a4 {+0.32/28 5.57} Qh2+ {0.00/32 6.55}
26. Kf1 {+0.47/27 4.32} Qh3+ {0.00/35 5.34}
27. Qg2 {+0.60/28 3.62} Qc8 {+0.21/28 5.32}
28. e3 {+1.09/23 3.17} Qg4 {+1.18/27 4.84}
29. f3 {-1.47/25 9.14} Qf5 {+1.44/28 18.85}
30. Ke2 {-1.55/25 8.08} c5 {+1.47/28 3.83}
31. g4 {-1.50/27 14.61} Qd7 {+1.87/26 9.39}
32. Rh1 {-1.50/25 10.33} cxd4 {+1.61/29 6.05}
33. cxd4 {-1.63/27 10.94} Rhe8 {+1.80/30 10.12}
34. Kf1 {-2.02/26 4.80} Qxa4 {+1.82/30 15.84}
35. Qf2 {-2.17/27 4.33} Qc4+ {+1.92/32 3.89}
36. Kg1 {-2.49/26 3.00} Qd3 {+1.73/33 14.45}
37. Rb2 {-1.70/25 3.00} Bc7 {+1.80/30 3.93}
38. Qf1 {-1.96/23 2.99} Rxd4 {+2.17/30 3.06}
39. Qxd3 {-2.80/24 2.99} Rxd3 {+2.33/31 6.04}
40. Kf2 {-2.94/26 3.01} Be5 {+2.15/30 2.89}
41. Ra2 {-2.79/25 2.99} b6 {+2.24/32 4.30}
42. Ke2 {-2.96/25 2.99} Rb3 {+2.69/33 13.05}
43. Rh7 {-3.39/24 2.83} a5 {+3.25/32 11.78}
44. Rd7 {-3.81/25 3.18} Rh8 {+3.34/32 2.41}
45. f4 {-4.06/26 3.02} gxf4 {+3.54/33 10.82}
46. exf4 {-4.31/25 2.99} Rh2+ {+4.03/31 8.87}
47. Kd1 {-4.57/27 3.01} Kc8 {+4.03/33 3.22}
48. Rd5 {-4.47/25 2.97} Bc7 {+4.29/32 4.98}
49. Rc2 {-4.71/23 3.00} Rg3 {+4.26/34 2.24}
50. Rc4 {-5.04/23 3.01} Rxg4 {+4.41/27 2.13}
51. Kc1 {-5.35/23 3.01} Rg7 {+4.98/29 8.11}
52. Rcd4 {-5.72/23 2.99} Re7 {+5.10/32 6.70}
53. Kb1 {-5.80/22 3.00} Kb7 {+5.36/30 7.78}
54. Bc3 {-6.07/23 3.01} Rf2 {+6.63/25 5.74}
55. Rd2 {-6.59/23 3.01} Rxf4 {+6.76/23 0.66}
56. R5d3 {-7.18/22 2.98} Ka6 {+7.06/23 1.96}
57. Rd1 {-7.65/22 3.02} Re2 {+7.20/24 1.54}
58. R1d2 {-9.32/23 3.00} Rxd2 {+7.65/27 4.93}
59. Bxd2 {-10.53/24 3.00} Re4 {+7.77/28 3.71}
60. Kc2 {-7.68/22 2.98} a4 {+7.99/28 2.53}
61. Rh3 {-9.80/23 3.01} b5 {+8.44/27 4.18}
62. Kd3 {-11.02/23 3.00} Rg4 {+9.32/27 4.49}
63. Rh7 {-11.62/22 2.99} Be5 {+10.13/25 2.43}
64. Be3 {-12.84/22 3.01} Ka5 {+10.13/27 1.38}
65. Kc2 {-13.76/22 3.01} Kb4 {+12.55/26 5.19}
66. Bc1 {-17.42/23 2.99} Kc4 {+14.66/29 3.00}
67. Rh8 {-15.67/24 2.98} Rg2+ {+9997.01/27 0.97}
68. Kd1 {-65.71/25 3.02}
{Black wins by adjudication}
0-1[/pgn]
[d]1kqr3r/ppp5/1b3p2/6p1/P2P4/2P3P1/3BPPQ1/1RR2K2 w - - 5 28

May be someone can confirm with his output for that position with the commercial Ethereal 13.25 NNUE version?

Krzysztof Grzelak · Post by **Krzysztof Grzelak** » Sat Jan 01, 2022 10:54 pm

Guenther wrote: ↑Sat Jan 01, 2022 10:33 pm
Krzysztof Grzelak wrote: ↑Sat Jan 01, 2022 7:11 pm So I wonder if the Ethereal program is well written by the programmer. Because I have the impression that the programmer has never tested his engine without a book.
A few things:
If you test from the same position always, you completely defeat the purpose of any statistical meaningful rating calculation.
The games are not independent, this just produces statistical noise then. Actually it is the same like playing just a few games instead.
This was mentioned dozens of times surely over the last 20 years in this forum, just do some research, I won't explain it here
again in detail.

But, I also noticed that something is very suspicious in your games. I took one example with a strange eval (depth also surprisingly low BTW)
and despite not having the commercial NNUE version here, my compilation of 13.43 HCE (not much different from 13.25 HCE wise) has nearly
the same evaluation than Ethereal NNUE in your game??? (on a slow hardware)
As long as I don't see any debug, which confirms you really use the nnue file, I would believe you didn't use it at all in your games...

Ethereal 13.43 SSSE3 GS
1kqr3r/ppp5/1b3p2/6p1/P2P4/2P3P1/3BPPQ1/1RR2K2 w - - 5 28
Code: Select all
dep	score	nodes	time	(not shown:  tbhits	knps	seldep)
 27	 -0.01 	70.1M  	1:35.70	b1a1 b6a5 a1b1 c7c6 e2e4 c8e6 f1g1 h8h7 d4d5 c6d5 e4d5 d8d5 d2e3 d5e5 e3d4 e5e4 d4f6 g5g4 f6d4 e4e2 b1b5 a5b6 c1b1 e2e1 b1e1 e6e1 g2f1 e1e4 f1g2 e4e2 d4b6 a7b6 b5b1 
 26	+1.06 	22.0M  	0:28.95	e2e3 h8h3 f1e2 c7c6 c3c4 c8e6 d4d5 c6d5 a4a5 b6c7 b1b5 e6g4 g2f3 g4f3 e2f3 h3h2 a5a6 b7b6 b5d5 g5g4 f3g4 h2f2 d5d8 c7d8 d2c3 f6f5 g4h3 f2f3 c3d4 d8c7 c1g1 c7d6 h3h4 b8c8 g3g4 f5g4 h4g4 f3f8 g1d1 d6c7 g4g5 f8g8 g5f6 
 26	+1.12 	18.4M  	0:24.08	e2e3 
 26	+1.17 	16.8M  	0:21.99	e2e3 
 25	+1.07 	15.9M  	0:20.88	e2e3 h8h3 f1e2 c7c6 c3c4 c8e6 d4d5 c6d5 a4a5 b6c7 a5a6 b7b6 b1b4 h3h7 d2c3 f6f5 c4d5 e6d5 g2d5 d8d5 b4c4 f5f4 g3f4 g5f4 e3e4 d5b5 c3f6 h7d7 f6d4 
 25	+1.21 	15.5M  	0:20.32	e2e3 
 25	+0.99 	14.9M  	0:19.54	e2e3 
 25	+1.04 	13.3M  	0:17.42	e2e3 
 24	+0.94 	12.6M  	0:16.42	e2e3 h8h3 f1e2 c7c6 c3c4 c8g4 g2f3 g4e6 e2d3 f6f5 a4a5 b6c7 a5a6 b7b6 c4c5 g5g4 f3g2 h3h6 c5b6 a7b6 b1b4 b8a7 d3e2 e6e4 g2e4 f5e4 b4a4 h6f6 a4b4 a7a6 
 24	+1.07 	12.3M  	0:16.00	e2e3 
 24	+1.11 	11.4M  	0:14.69	e2e3 
 24	+1.16 	10.5M  	0:13.61	e2e3 
 23	+1.06 	9.71M  	0:12.46	e2e3 h8h3 f1e2 c7c6 c3c4 c8g4 g2f3 g4e6 e2d3 f6f5 a4a5 g5g4 f3g2 b6c5 c1h1 c5f8 h1h3 g4h3 g2f3 f8g7 b1h1 d8h8 a5a6 b7b6 d3c3 e6e4 f3e4 f5e4 c3c2 
 23	+1.23 	9.43M  	0:12.10	e2e3 
 23	+1.88 	5.01M  	0:06.19	e2e3 
 22	+1.78 	4.22M  	0:05.21	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b1 b8c8 f1g2 g5g4 c1h1 f6f5 h1h4 h8h4 g3h4 d8h8 b1h1 h8h5 d4c5 b6c5 g2g3 c8d7 g3f4 d7e6 a4a5 
 22	+1.84 	4.11M  	0:05.07	e2e3 
 21	+1.94 	2.74M  	0:03.33	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b1 b8c8 f1g2 g5g4 c1h1 b6c7 b1e1 f6f5 e1b1 c7b6 h1h4 h8h4 g3h4 d8h8 b1h1 h8h7 d4c5 b6c5 g2g3 h7b7 
 21	+1.93 	2.45M  	0:02.99	e2e3 
 20	+1.83 	2.16M  	0:02.63	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b1 b8c8 f1g2 g5g4 c1h1 b6c7 b1e1 f6f5 e1b1 c5c4 h1h8 d8h8 b1b4 c8d7 b4b7 
 20	+1.89 	2.11M  	0:02.57	e2e3 
 19	+1.84 	1.43M  	0:01.73	e2e3 c7c5 b1b5 c8c6 g2c6 b7c6 b5b3 b8a8 f1g2 g5g4 b3b1 b6c7 e3e4 c5d4 c3d4 d8d4 c1c6 d4d2 c6c7 
 18	+1.68 	854223	0:01.04	e2e3 c8f5 d2e1 d8e8 a4a5 f5a5 c3c4 a5f5 c4c5 e8e3 c5b6 a7b6 f1g1 e3f3 e1d2 f5g4 b1a1 g4f5 d2e3 c7c6 c1b1 
 17	+1.24 	682086	0:00.84	e2e3 c8f5 d2e1 d8e8 a4a5 f5a5 c3c4 a5f5 c4c5 e8e3 c5b6 a7b6 f1g1 e3a3 b1a1 a3d3 a1a4 d3f3 e1d2 f5d3 
 16	+1.36 	410613	0:00.51	e2e3 c8f5 d2e1 h8e8 f1g1 c7c6 g3g4 f5e6 c3c4 b6c7 b1b2 b8a8 c1b1 c7b6 a4a5 
 15	+1.33 	243352	0:00.31	e2e3 c7c5 b1b5 c8e6 f1e1 e6c4 c1b1 c5d4 c3d4 c4a4 g2f3 b8a8 b5f5 a4d7 f5f6 
 14	+1.01 	201245	0:00.26	e2e3 c7c5 b1b5 c8e6 f1e1 e6c4 c1b1 c5d4 c3d4 c4a4 g2f3 d8f8 
 13	+1.11 	70507  	0:00.09	e2e3 c7c5 b1b5 c8e6 f1e1 e6d7 c1b1 d7c6 g2c6 b7c6 b5b6 a7b6 b1b6 b8a7 b6c6 
 12	+0.98 	45843  	0:00.06	e2e3 c7c5 b1b5 c8e6 c1b1 e6c4 f1e1 c5d4 c3d4 c4a4 b5b6 a7b6 b1b6 
 11	+0.69 	32285  	0:00.04	e2e3 c7c5 b1b5 c8f5 c1b1 f5d3 f1e1 d3f5 
 10	+0.26 	29070  	0:00.04	e2e3 c7c5 b1b5 c8f5 c1b1 f5d3 f1e1 c5d4 c3d4 
  9	+0.44 	23260  	0:00.03	e2e3 c7c6 c3c4 c8e6 f1e2 d8e8 c4c5 
  8	+0.45 	12180  	0:00.01	c1d1 c8f5 b1b5 f5e6 
[pgn]
[Event "My Tournament"]
[Site "?"]
[Date "2021.12.30"]
[Round "36"]
[White "Ethereal 13.25 x64"]
[Black "Revenge 2.0 x64"]
[Result "0-1"]
[ECO "A20"]
[GameDuration "00:16:44"]
[GameEndTime "2021-12-30T18:38:05.686 Œrodkowoeuropejski czas stand."]
[GameStartTime "2021-12-30T18:21:21.173 Œrodkowoeuropejski czas stand."]
[Opening "English Opening"]
[PlyCount "135"]
[Termination "adjudication"]
[TimeControl "300+3"]

1. c4 {+0.30/25 19.46} e5 {-0.19/26 8.27}
2. g3 {+0.35/25 8.27} Nf6 {-0.19/27 9.56}
3. Bg2 {+0.29/25 13.65} d5 {-0.19/29 9.06}
4. cxd5 {+0.46/26 9.25} Nxd5 {-0.19/30 8.94}
5. Nf3 {+0.38/26 7.63} Nc6 {-0.19/30 29.27}
6. O-O {+0.47/27 8.76} e4 {-0.17/28 11.85}
7. Ne1 {+0.92/28 16.64} h5 {-0.29/27 9.92}
8. Qa4 {+0.89/27 31.23} Be6 {0.00/27 6.47}
9. Qxe4 {+0.87/29 30.36} Qe7 {0.00/27 10.81}
10. d3 {+0.78/30 44.94} h4 {0.00/27 6.14}
11. Nf3 {+0.42/29 14.75} hxg3 {0.00/28 6.00}
12. hxg3 {+0.45/30 20.89} O-O-O {0.00/31 23.18}
13. Nc3 {+0.41/27 6.61} f6 {0.00/32 9.45}
14. Bd2 {+0.01/26 10.30} Kb8 {0.00/29 6.44}
15. Rfc1 {+0.56/27 6.11} Qd7 {-0.16/30 23.65}
16. Nd4 {+0.66/27 5.45} Nxd4 {0.00/35 5.22}
17. Qxd4 {+0.73/26 3.86} Nxc3 {0.00/35 6.47}
18. bxc3 {+0.48/28 7.41} Qc8 {-0.07/34 21.74}
19. Qa4 {+0.40/27 8.91} Bc5 {0.00/29 5.21}
20. Rab1 {+0.01/28 13.51} Bd7 {0.00/32 6.30}
21. Qf4 {+0.18/31 12.94} g5 {0.00/34 6.30}
22. Qf3 {+0.16/30 9.46} Bb6 {0.00/35 5.01}
23. d4 {+0.01/30 9.44} Bh3 {0.00/33 6.13}
24. Bxh3 {+0.32/29 5.79} Qxh3 {0.00/32 8.02}
25. a4 {+0.32/28 5.57} Qh2+ {0.00/32 6.55}
26. Kf1 {+0.47/27 4.32} Qh3+ {0.00/35 5.34}
27. Qg2 {+0.60/28 3.62} Qc8 {+0.21/28 5.32}
28. e3 {+1.09/23 3.17} Qg4 {+1.18/27 4.84}
29. f3 {-1.47/25 9.14} Qf5 {+1.44/28 18.85}
30. Ke2 {-1.55/25 8.08} c5 {+1.47/28 3.83}
31. g4 {-1.50/27 14.61} Qd7 {+1.87/26 9.39}
32. Rh1 {-1.50/25 10.33} cxd4 {+1.61/29 6.05}
33. cxd4 {-1.63/27 10.94} Rhe8 {+1.80/30 10.12}
34. Kf1 {-2.02/26 4.80} Qxa4 {+1.82/30 15.84}
35. Qf2 {-2.17/27 4.33} Qc4+ {+1.92/32 3.89}
36. Kg1 {-2.49/26 3.00} Qd3 {+1.73/33 14.45}
37. Rb2 {-1.70/25 3.00} Bc7 {+1.80/30 3.93}
38. Qf1 {-1.96/23 2.99} Rxd4 {+2.17/30 3.06}
39. Qxd3 {-2.80/24 2.99} Rxd3 {+2.33/31 6.04}
40. Kf2 {-2.94/26 3.01} Be5 {+2.15/30 2.89}
41. Ra2 {-2.79/25 2.99} b6 {+2.24/32 4.30}
42. Ke2 {-2.96/25 2.99} Rb3 {+2.69/33 13.05}
43. Rh7 {-3.39/24 2.83} a5 {+3.25/32 11.78}
44. Rd7 {-3.81/25 3.18} Rh8 {+3.34/32 2.41}
45. f4 {-4.06/26 3.02} gxf4 {+3.54/33 10.82}
46. exf4 {-4.31/25 2.99} Rh2+ {+4.03/31 8.87}
47. Kd1 {-4.57/27 3.01} Kc8 {+4.03/33 3.22}
48. Rd5 {-4.47/25 2.97} Bc7 {+4.29/32 4.98}
49. Rc2 {-4.71/23 3.00} Rg3 {+4.26/34 2.24}
50. Rc4 {-5.04/23 3.01} Rxg4 {+4.41/27 2.13}
51. Kc1 {-5.35/23 3.01} Rg7 {+4.98/29 8.11}
52. Rcd4 {-5.72/23 2.99} Re7 {+5.10/32 6.70}
53. Kb1 {-5.80/22 3.00} Kb7 {+5.36/30 7.78}
54. Bc3 {-6.07/23 3.01} Rf2 {+6.63/25 5.74}
55. Rd2 {-6.59/23 3.01} Rxf4 {+6.76/23 0.66}
56. R5d3 {-7.18/22 2.98} Ka6 {+7.06/23 1.96}
57. Rd1 {-7.65/22 3.02} Re2 {+7.20/24 1.54}
58. R1d2 {-9.32/23 3.00} Rxd2 {+7.65/27 4.93}
59. Bxd2 {-10.53/24 3.00} Re4 {+7.77/28 3.71}
60. Kc2 {-7.68/22 2.98} a4 {+7.99/28 2.53}
61. Rh3 {-9.80/23 3.01} b5 {+8.44/27 4.18}
62. Kd3 {-11.02/23 3.00} Rg4 {+9.32/27 4.49}
63. Rh7 {-11.62/22 2.99} Be5 {+10.13/25 2.43}
64. Be3 {-12.84/22 3.01} Ka5 {+10.13/27 1.38}
65. Kc2 {-13.76/22 3.01} Kb4 {+12.55/26 5.19}
66. Bc1 {-17.42/23 2.99} Kc4 {+14.66/29 3.00}
67. Rh8 {-15.67/24 2.98} Rg2+ {+9997.01/27 0.97}
68. Kd1 {-65.71/25 3.02}
{Black wins by adjudication}
0-1[/pgn]
[d]1kqr3r/ppp5/1b3p2/6p1/P2P4/2P3P1/3BPPQ1/1RR2K2 w - - 5 28

May be someone can confirm with his output for that position with the commercial Ethereal 13.25 NNUE version?

I'm sorry, but I don't have to lie to you, Guenther. The engine is set like this.

Ethereal - question

Ethereal - question

Re: Ethereal - question

Re: Ethereal - question

Re: Ethereal - question

Re: Ethereal - question

Re: Ethereal - question

Re: Ethereal - question

Re: Ethereal - question

Re: Ethereal - question

Re: Ethereal - question