EN Engine Test 2023

[Eduard]

EN Engine Test 2023 (ENET-2023

090323: Hello friends,
today is the start of my new engine test. The test contains 110 positions. Many of these are from the EN-Test 2022, because I like them. Many new positions have been added. I removed some positions that were easy to solve.

I already posted most of the new positions here a few weeks ago:
https://talkchess.com/forum3/viewtopic. ... 70#p942947

Download EN-2023-Test (CBH & PGN) Pixeldrain & Filehorst.de
https://solistachess.jimdosite.com/testing/

(Maybe someone can convert the positions to EPD, that would be nice)

Before I start a new engine test run, I completely shut down the computer for 3 to 5 minutes. Each engine has only one test attempt. I know that multiple attempts may produce different results. This is not tragic, because I intend to test new versions more often. Example Stockfish. There is at least one new developer release every week. I don't know if I will test every version, but I will test as many as possible. If I test Stockfish several times a month, that should be enough. The best version always remains in the list. A worse result is not taken into account. I will not test engines that are older than 6 months. The first engine I tested is Stockfish dev. Here is the result.

AMD Ryzen 3900X, 20 Threads, Hash 4 GB, all 3456men Syzygy. Timecontrol 60s:

1) Stockfish dev 080323, Result: 91 out of 110 = 82.7%. Stockfish dev.txt (ZIP)

I'll post the result here in full so that you can get an overview of the 110 positions. New results can be downloaded as TXT files from my homepage.

Ryzen 3900X, 20 Threads, hash 4 GB, 3456men Syzygy. Stockfish dev 080323:



Solving: E:\Database\ENET-2023.cbh
Maximum time = 60s.

1. ENET-2023 001 (ERET 001) - Entlastung, Rf4, Akopian-Khenkin, Tilburg 1994 Solved in 0.63s/14; Solved: 1
2. ENET-2023 002 (ERET 002) - Zugzwang, Ng6+, Akerblom 1961 Solved in 37.14s/30; Solved: 2
3. ENET-2023 003 (ERET 003) - Linienoeffnen, f5, Alvarez-Kokkila, CorrGame 1997 Solved in 0.14s/10; Solved: 3
4. ENET-2023 004 (ERET 004) - Endspiel L&S vs L&S, Nxg5, Anand-Shirov, Advanced Chess 2000 Solved in 0.19s/24; Solved: 4
5. ENET-2023 005 (ERET 005) - Laeuferopfer f7, Bxf7+, Andriulaitis-Ruefenacht Solved in 0.06s/10; Solved: 5
6. ENET-2023 006 (ERET 009) - Springeropfer, Ng6, Aseev-Popv, Kazan 1995 Solved in 0.06s/12; Solved: 6
7. ENET-2023 007 (HTC108),2 - Damenopfer, Qxf6 Solved in 5.45s/28; Solved: 7
8. ENET-2023 008 (ERET 013) - Zentrum, Qd2, Bang-Umansky, CorrGame 1996 Solved in 2.56s/27; Solved: 8
9. ENET-2023 009 (Taktik) - Turmopfer, Rd6 Solved in 42.47s/41; Solved: 9
10. ENET-2023 010 (ERET 015) - Endspiel D&B vs T&L&B, Qd3+, Becker 2015 Solved in 14.05s/63; Solved: 10
11. ENET-2023 011 (HTC10) - Caruana vs Carlsen, g4 > 60s.
12. ENET-2023 012 (ERET 019) - Endspiel L&L&S&S vs L&S&B, Nf5, Broni 1970 Solved in 13.72s/20; Solved: 11
13. ENET-2023 013 (ERET 020) - Laeuferopfer h7, Bxh7, Burmakin-Ziatdinov, Moskau 2009 Solved in 0.06s/10; Solved: 12
14. ENET-2023 014 (ACT-Suite) - Turmopfer, Rxg7 Solved in 3.05s/27; Solved: 13
15. ENET-2023 015 (ERET 023) - Rochadeangriff, Ng6+, Copie-Patrici, CorrGame 1986 Solved in 0.05s/10; Solved: 14
16. ENET-2023 016 (HTC108) - Springeropfer, Nc4 Solved in 10.72s/44; Solved: 15
17. ENET-2023 017 (ERET 025) - Endspiel T&L vs T&L, Bxd8, Critter-Naum, CompGame 2015 Solved in 2.98s/38; Solved: 16
18. ENET-2023 018 (ERET 027) - Zugzwang, g4, DeLimburg 1900 Solved in 0.03s/10; Solved: 17
19. ENET-2023 019 (CCC) - Attack, Bxh7+ Solved in 5.08s/22; Solved: 18
20. ENET-2023 020 (ERET 030) - Initiative, Nh4, Dutra-Alfredo, CorrGame 2008 Solved in 0.06s/13; Solved: 19
21. ENET-2023 021 (ERET 031) - Endspiel L vs L, b4, Estrin-Ivashin, CorrGame 1947 Solved in 0.03s/10; Solved: 20
22. ENET-2023 022 (CCC) - Attack, Nh4 > 60s.
23. ENET-2023 023 (HTC108-21) - Manoever, Kc1 Solved in 4.14s/25; Solved: 21
24. ENET-2023 024 (Gashimov) - Springeropfer, Ng6+ > 60s.
25. ENET-2023 025 (CCC) - Ausgleich, Rf4 Solved in 58.55s/31; Solved: 22
26. ENET-2023 026 (ERET 038) - Endspiel S&S vs L&B, exd5, Garcia 2000 Solved in 0.36s/18; Solved: 23
27. ENET-2023 027 (Bondarenko) - Endspiel, Bf4 > 60s.
28. ENET-2023 028 (ERET 041) - Springeropfer, Nxg5, Gerber-Georgiev, Luzern 2003 Solved in 0.20s/10; Solved: 24
29. ENET-2023 029 (ERET 043) - Springeropfer, Nxf5, Glek-Ulibin, Wien 1998 Solved in 0.02s/10; Solved: 25
30. ENET-2023 030 (Gusev) - Damenopfer, Qxe5 > 60s.
31. ENET-2023 031 (HTC108) - Laeferopfer, Bxc5+ > 60s.
32. ENET-2023 032 (ERET 046) - Endspiel T&S vs T&S, Nd3, Grandelius-Raznikov, Albena 2011 Solved in 25.84s/54; Solved: 26
33. ENET-2023 033 (ERET 050) - Koenigsangriff, Nf3+, Hansen-Barua, Biel 1993 Solved in 0.81s/20; Solved: 27
34. ENET-2023 034 (Onlinegame) - Lc0 vs Stockfish, f4! Lc0 vs Stockfish, Onlinegame Solved in 0.08s/13; Solved: 28
35. ENET-2023 035 (ERET 054) - Endspiel D&T&L vs D&T&L, c3, Jussupow-Kasparov, Linares 1990 Solved in 0.31s/20; Solved: 29
36. ENET-2023 036 (ERET 055) - Laeuferopfer h6, Bxh6, Kamsky-Lautier, Dortmund 1993 Solved in 0.05s/10; Solved: 30
37. ENET-2023 037 (ERET 056) - Zwischenzug, Be2, Karjakin-Kramnik, Moskau 2010 Solved in 5.95s/29; Solved: 31
38. ENET-2023 038 (ERET 059) - Endspiel D&L vs D&T, Qd3, Komodo-Stockfish, CompGame 2014 Solved in 0.11s/14; Solved: 32
39. ENET-2023 039 (Morozevich A) - Springeropfer, Nd5 > 60s.
40. ENET-2023 040 (ERET 061) - Koenigsangriff, Rf6, Kortschnoj-Shirov, Madrid 1996 Solved in 0.08s/15; Solved: 33
41. ENET-2023 041 (2015_Glarean) - Endspiel, Rg7+ > 60s.
42. ENET-2023 042 (Zepler,Studie) - Endspiel, g6 Solved in 26.06s/29; Solved: 34
43. ENET-2023 043 (Lasker) - Endspiel, Ke4 Solved in 20.27s/39; Solved: 35
44. ENET-2023 044 (ERET 065) - Zwischenzug, Qb6, Kuligowski-Knaak, Slupsk 1982 Solved in 0.69s/17; Solved: 36
45. ENET-2023 045 (ERET 066) - Endspiel T&S vs L&S, Ke1, Marwitz 1965 Solved in 1.41s/33; Solved: 37
46. ENET-2023 046 (ERET 069) - Koenigsangriff, h5, Morozevich-Petursson, London 1994 Solved in 0.11s/10; Solved: 38
47. ENET 2023 047 (ERET 070) - Starke Felder, a3, Motylev -Li, China 2010 Solved in 0.05s/10; Solved: 39
48. ENET-2023 048 (ERET 071) - Ablenkung, f5, Myers-Baufays, CorrGame 2010 Solved in 0.03s/10; Solved: 40
49. ENET-2023 049 (ERET 072) - Zentralisierung, Qd5, N.N.-N.N. (Analyse) Solved in 0.05s/10; Solved: 41
50. ENET-2023 050 (Los) - Springer-Manoever, Ne2 Solved in 6.33s/31; Solved: 42
51. ENET-2023 051 (Onlinegame) - PurePower vs Exxon, Qa4, Online Engine game Solved in 3.66s/29; Solved: 43
52. ENET-2023 052 (ERET 075) - Festung, Qf5, N.N.-N.N. (Analyse) Solved in 0.41s/18; Solved: 44
53. ENET-2023 053 (Bauer Robert) - Verteidigung, g3 Solved in 0.42s/19; Solved: 45
54. ENET-2023 054 (Mihai ENeghina - Gegenangriff, Nxb5 Solved in 54.28s/35; Solved: 46
55. ENET-2023 055 (Tauber - Sir. - Druck, e5 Solved in 44.31s/37; Solved: 47
56. ENET-2023 056 (ERET 080) - Raeumung, Bc1, Nirvana-Hannibal, CompGame 2015 Solved in 0.64s/20; Solved: 48
57. ENET-2023 057 (ERET 081) - Koenigsangriff, Bg7, Nunn-Ward, GB 1998 Solved in 0.02s/10; Solved: 49
58. ENET-2023 058 (TCEC Nr.8 j329 - Endspiel, Kf4 Solved in 5.33s/29; Solved: 50
59. ENET-2023 059 (ERET 083) - Abtausch, Nb5, Orlov-Georgiev, Belgrad 1988 Solved in 0.08s/10; Solved: 51
60. ENET-2023 060 (Sikorsky Horst - Mittelspiel, Qe1 Solved in 24.34s/35; Solved: 52
61. ENET-2023 061 (ERET 085) - Abtausch, Ng4, Petrosian-Bronstein, UdSSR-Ch 1967 Solved in 0.09s/14; Solved: 53
62. ENET-2023 062 (ERET 086) - Qualitaetsopfer, c5, Petrosian-Fischer, BuenosAires 1971 Solved in 0.08s/10; Solved: 54
63. ENET-2023 063 (ERET 087) - Raeumung, e5, Petrosian-Larsen, PalmaDeMallorca Solved in 0.20s/15; Solved: 55
64. ENET-2023 064 (ERET 088) - Endspiel S vs S, g4, Razuvaev-Ostojic, 1988 Solved in 0.05s/10; Solved: 56
65. ENET-2023 065 (ERET 089) - Unterverwandlung, e8N, Rumaenien 1977 Solved in 10.92s/22; Solved: 57
66. ENET-2023 066 (ERET 090) - Endspiel L vs L, f4, Salai 2011 > 60s.
67. ENET 067 (Study) - Endspiel, Rc2 Solved in 39.72s/64; Solved: 58
68. ENET-2023 068 (HHdbVI) - Angriff, Qa7+ Solved in 0.44s/24; Solved: 59
69. ENET-2023 069 (TTT2.053) - Springeropfer, Ng6 > 60s.
70. ENET-2023 070 (ERET 094) - Damenopfer, Rf6, Sirobaba-Chaika, CorrGame 2011 > 60s.
71. ENET-2023 071 (ERET 096) - Endspiel D vs T&L&S, Kf6, Sochnev 2009 Solved in 9.22s/54; Solved: 60
72. ENET-2023 072 (ERET 097) - Endspiel L vs L, c5, Spassky-Barne, SanJuan 1974 Solved in 0.59s/25; Solved: 61
73. ENET-2023 073 (Smyslov) - Endspiel, g4 > 60s.
74. ENET-2023 074 - Nemeth vs Fritz, Blitzgame Solved in 0.25s/11; Solved: 62
75. ENET-2023 075 - AlphaZero vs Stockfish, Bg5 Solved in 0.22s/17; Solved: 63
76. ENET-2023 076 (ERET 104) - Entwicklung, Ne7, VanOosterom-Timmerman, CorrGame Solved in 0.03s/10; Solved: 64
77. ENET-2023 077 - Ollinegame beste Verteidigung, Qb7 Solved in 0.45s/19; Solved: 65
78. ENET-2023 078 - Mittelspiel, Re4 Solved in 28.98s/35; Solved: 66
79. ENET-2023 079 - Onlinegame Stockfish vs Lc0, Na6 Solved in 3.53s/50; Solved: 67
80. ENET-2023 080 (ERET 110) - Freibauer, c5, Ye-Lutz, Batumi 2001 Solved in 1.77s/21; Solved: 68
81. ENET-2023 081 - Onlinegame Beste Verteidigung, Qxf6! Solved in 3.11s/30; Solved: 69
82. ENET-2023 082 - Onlinegame, Bxf5 (dxc3?) Solved in 0.88s/24; Solved: 70
83. ENET-2023 083 - Lc0 vs Stockfish, Onlinegame, b4! Solved in 49.39s/59; Solved: 71
84. ENET-2023 084 - Onlinegame Best move, Rfc8 Solved in 9.30s/26; Solved: 72
85. ENET-2023 085 - Onlinegame, g4! Solved in 13.42s/35; Solved: 73
86. ENET-2023 086 - Online Engine game Solista, f6! > 60s.
87. ENET-2023 087 - Online engine game, Bxh7+! Solved in 10.08s/36; Solved: 74
88. ENET-2023 088 - Dorsz, Stockfish v Blue Marli, Qh6! > 60s.
89. ENET-2023 089 - Taktik, Verteidigung, Qe7 > 60s.
90. ENET-2023 090 - Zugzwang, f6! Solved in 25.75s/55; Solved: 75
91. ENET-2023 091 - Turmopfer, Rxf3! Solved in 17.89s/33; Solved: 76
92. ENET-2023 092 - Angriff, Kf2 Solved in 18.73s/31; Solved: 77
93. ENET-2023 093 - Königsangriff, h5! > 60s.
94. ENET-2023 094 - Mattopfer, Qc6 Solved in 7.63s/25; Solved: 78
95. ENET-2023 095 - Läuferopfer, Bf6 Solved in 6.39s/32; Solved: 79
96. ENET-2023 096 - Läuferopfer, Bxf7, Carlsen Solved in 8.70s/31; Solved: 80
97. ENET-2023 097 - Endspiel, Rf1! Solved in 43.52s/37; Solved: 81
98. ENET-2023 098 - Linieöffnung, Ng5! Solved in 12.03s/29; Solved: 82
99. ENET-2023 099 - Endspiel, f5! Solved in 8.33s/49; Solved: 83
100. ENET-2023 100 - Mittelspiel, Re8! > 60s.
101. ENET-2023 101 - Damenausfall, Qh5! > 60s.
102. ENET-2023 102 - Königsangriff, Qa5! Solved in 0.63s/20; Solved: 84
103. ENET-2023 103 - Angriff, Ra3! Solved in 19.19s/31; Solved: 85
104. ENET-2023 104 - Spieloeffnung, h4! Solved in 3.05s/26; Solved: 86
105. ENET-2023 105 - Wichtiger Abtausch, Nxc3! Solved in 1.44s/23; Solved: 87
106. ENET-2023 106 - Mittelspiel Entlastung, Ne4! Solved in 3.92s/34; Solved: 88
107. ENET-2023 107 - Mittellspiel, Rfe8! Solved in 0.78s/23; Solved: 89
108. ENET-2023 108 - Bestmove, Rad1! Kramnik vs Howell Solved in 0.30s/18; Solved: 90
109. ENET-2023 109 - Stockfish vs Komodo, Bd3, Bestmove Solved in 0.19s/15; Solved: 91
110. ENET-2023 110 - Entlastung, Nd5 > 60s.

Result: 91 out of 110 = 82.7%. Average time = 8.62s / 25.12

[criko]

And soon we'll see all meaningless Nemeth engines on the 1st places

And the whole SF dev team has no clue of programming.

You're such a clown!

[Eduard]

I posted the new positions here a few weeks ago. Before that I had checked about 1000 positions used by Peter Martan (3 test suites). Many of these are also included in my test. Most of them, however, didn't make sense to me because they were too easy, foreign to practice, or buggy. Unfortunately, I had to delete 3 positions from my old test because they contained secondary solutions.

The test is not tuned to any engine, and Stockfish dev was the first engine to pass this test. I have a clear conscience, everything happened as I wrote.

But why should it be wrong to also tune an engine with position tests? Tests like this have been around for decades. I remember M. Gurevich in the CSS forum. Gurevich had a tester (what was his name?) who did an engine ranking.

Walter Eigenmann also created his ERET test a few years ago, which was very useful.

I sent my first engine to two testers, K. Utzinger and D. Uter. I'm not testing myself at this point. The result of D. Uter is on my homepage, and I posted the results of K. Utzinger in the chess2u forum.

[RubiChess]

But why should it be wrong to also tune an engine with position tests?

Find the reason here: https://github.com/Matthies/RubiChess/w ... uter-chess

[Eduard]

On my homepage there is now also a download in EPD.

[Eduard]

But why should it be wrong to also tune an engine with position tests?

Find the reason here: https://github.com/Matthies/RubiChess/w ... uter-chess

Of course,
if an engine is only based on position tests, that's not enough. Me and my friends play thousands of games live on servers and also play tournaments.

I always get the latest games sent to me. I check and analyze every game. That's great fun for me. When I lose, I also study the engine's moves and try to improve some things as far as my modest skills go.

Recently Leptir Analyzer (not made for games) took first place twice on playchess with only 2700 kns. That may be coincidence, but I believe that the engine can also do something. So I must have done something right. In position tests (not only mine) it is one of the best engines.

If you're all so smart, do better, thanks.

[Eduard]

https://solistachess.jimdosite.com/testing/

280323:
I noticed that ShashChess 30 has deteriorated by 2 points compared to v29.1. Why is that? ShashChess 30 uses a different NNUE network than version 29.1, a network not used by Stockfish dev, at the time of this note. Eman 8.90 has also deteriorated significantly (7 points) compared to Eman 8.81. Unfortunately, it is also not possible to say which NNUE network Eman 8.90 is currently using, because all NNUE networks at Eman have been renamed and are called "eval.nnue". Lc0 v30.0 has been tested with the BT2 network on RTX 3080ti, currently leading the TCEC S24 - Division P at the time of this notice. The time control at TCEC is 60'+6" and it is played on super strong hardware!

GUI Powerfritz 18, Ryzen 3900X, 20 Threads, Hash 4 GB, all 3456men Syzygy. Time 60s for move:

1) Leptir Analyzer, Result: 103 out of 110 = 93.6%. Leptir Analyzer.txt (ZIP)
2) Crystal 5 KWK, Result: 100 out of 110 = 90.9%. Crystal 5 KWK.txt (ZIP)
3-4) Corchess 3 010323, Result: 94 out of 110 = 85.4%. Corchess 3 010323.txt (ZIP)
3-4) Blue Marlin 15.7, Result: 94 out of 110 = 85.4%. Blue Marlin 15.7.txt (ZIP)
5) Stockfish dev 080323, Result: 91 out of 110 = 82.7%. Stockfish dev.txt (ZIP)
6) Dark SisTer 4.6, Result: 90 out of 110 = 81.8%. Dark SisTer 4.6.txt (ZIP)
7) ShashChess 30, Result: 89 out of 110 = 80.9%. ShashChess 30.txt (ZIP)
8) Eman 8.90, Result: 88 out of 110 = 80.0%. Eman 8.90.txt (ZIP)
9) Lc0 v0.30.0-BT2 RTX 3080ti, Total: 75 (68%). Lc0 v0.30.0-BT2 RTX 3080ti.txt (ZIP)
10) Rebel-16.2, Result: 50 out of 110 = 45.4%. Rebel-16.2.txt (ZIP)
11) Powerfritz 18, Result: 45 out of 110 = 40.9%. Powerfritz 18.txt (ZIP)

Download all txt files on my homepage.

[CornfedForever]

https://solistachess.jimdosite.com/testing/

280323:
I noticed ...

My how quickly... "a few weeks" passes. A sure sign I'm getting old.

[Jouni]

Lc0 v30.0 has been tested with the BT2 network on RTX 3080ti, currently leading the TCEC S24 - Division P at the time of this notice?? NO SF leading 28.3.

[Eduard]

The test was now carried out with 5 different GPU configurations.
The best configuration is 2 GPUs connected in parallel. With 3 GPU that became
result weaker.

The RTX 3060 Ti has the best price/performance ratio.

1) Leptir Analyzer, Result: 103 out of 110 = 93.6%. Leptir Analyzer.txt (ZIP)
2) Crystal 5 KWK, Result: 100 out of 110 = 90.9%. Crystal 5 KWK.txt (ZIP)
3-4) Corchess 3 010323, Result: 94 out of 110 = 85.4%. Corchess 3 010323.txt (ZIP)
3-4) Blue Marlin 15.7, Result: 94 out of 110 = 85.4%. Blue Marlin 15.7.txt (ZIP)
5) Stockfish dev 080323, Result: 91 out of 110 = 82.7%. Stockfish dev.txt (ZIP)
6) Dark SisTer 4.6, Result: 90 out of 110 = 81.8%. Dark SisTer 4.6.txt (ZIP)
7) ShashChess 30, Result: 89 out of 110 = 80.9%. ShashChess 30.txt (ZIP)
8) Eman 8.90, Result: 88 out of 110 = 80.0%. Eman 8.90.txt (ZIP)
9) Lc0 v0.30 BT2, RTX 3080+3080ti, 86/110. Lc0 v0.30 BT2, 3080+3080ti.txt (ZIP)
10) Lc0 v0.30 BT2, RTX 4070ti, 80 aus 110 = 72.7%. Lc0 v0.30 BT2.txt (ZIP)
11) Lc0 v0.30 BT2 RTX 3080ti, Total: 75 (68%). Lc0 v0.30.0-BT2 RTX 3080ti.txt (ZIP)
12) Lc0 v0.30 BT2, RTX 3060m ti, 73 aus 110 = 66.3%. Lc0 v0.30 BT2.txt (ZIP)
13) Rebel-16.2, Result: 50 out of 110 = 45.4%. Rebel-16.2.txt (ZIP)
14) Powerfritz 18, Result: 45 out of 110 = 40.9%. Powerfritz 18.txt (ZIP)

TXT Files on my Homepage
https://solistachess.jimdosite.com/testing/