results of stockfish 100k opening position no book

[Uri Blass]

4 cores for every engine
1024 mbytes hash for every engine.
stockfish 100k nodes per move
Dragon 4000k or 40000k nodes per move
KomodoDragon 4M-Stockfish100k 40-10(+31 =18 -1)
KomodoDragon 40M-Stockfish100k 35-5(+30 =10)

I wonder if there is an engine that can get 100% or close to 100% against Stockfish100k in these conditions
Maybe changing the setting of KomdoDragon can help.
I used the default parameters and did not try to increase contempt.

[Uri Blass]

More nodes against stockfish wihtout changing setting seems not to help dragon to get significantly closer to 100%

KomodoDragon100M-Stockfish100k is only +6 =4 -0 for Dragon.
stockfish could even draw one game with the black pieces.

[Uri Blass]

Dragon3Contempt50 100M nodes per move-Stockfish100k nodes per move +7 =3 -0

[Uri Blass]

Stockfish could draw 3 games out of 5 games with the white pieces
last game of the match

I wonder if top humans know endgames so well to evaluate that black has only small winning chances in the following position and it is probably a draw.

[fen]1r4k1/5p2/4p1pp/8/2p1P3/b1B3PP/3R1P1K/8 b - - 0 42 [/fen]

[pgn][Event "Computer event"]
[Site "Somewhere on Earth"]
[Date "2022.05.26"]
[Round "1"]
[White "Stockfish 15"]
[Black "Dragon 3 by Komodo Chess 64-bit"]
[Result "1/2-1/2"]
[BlackElo "2000"]
[Time "06:27:41"]
[WhiteElo "2000"]
[TimeControl "600"]
[Termination "normal"]
[PlyCount "243"]
[WhiteType "human"]
[BlackType "program"]

1. e4 {100397} c5 {99419431} 2. Nf3 {100108} d6 {98828856} 3. Bb5+ {100311}
Bd7 {99033552} 4. Bxd7+ {+0.12/12 100318; B52: Sicilian, Canal-Sokolsky
attack, 3...Bd7} Qxd7 {98461825} 5. O-O {100501} Nf6 {98698717} 6. Re1
{100544} e6 {99307150} 7. c4 {100086} Be7 {99478073} 8. Nc3 {100245} O-O
{98791413} 9. d4 {100110} cxd4 {99511622} 10. Nxd4 {100075} a6 {99449699}
11. b3 {100215} Rc8 {98558121} 12. Bb2 {100309} Nc6 {98972232} 13. Na4
{100310} Qc7 {99682615} 14. Qd2 {100419} h6 {99462405} 15. Rac1 {100375}
Qd8 {98167833} 16. Rcd1 {100121} Nd7 {99797924} 17. Qe2 {100197} Qa5
{99462350} 18. Bc3 {100039} Qc7 {99386243} 19. h3 {100116} Rab8 {99579516}
20. Nxc6 {100380} Qxc6 {100000000} 21. Nb2 {100531} b5 {99660366} 22. Bd2
{100184} Rd8 {99683243} 23. cxb5 {100213} axb5 {99118459} 24. Nd3 {100421}
Rbc8 {100000000} 25. Ba5 {100121} Re8 {99665155} 26. Bb4 {100049} Qb6
{97949708} 27. Qe3 {100132} Qb7 {99529736} 28. a3 {100184} Rc2 {99976592}
29. Qf3 {100196} Nc5 {98763504} 30. Re3 {100249} Bg5 {100000000} 31. Nxc5
{100056} dxc5 {99476518} 32. Rc3 {100097} Rxc3 {98606711} 33. Bxc3 {100152}
Be7 {100000000} 34. Rd3 {100023} c4 {100000000} 35. Qg4 {100385} g6
{98183075} 36. bxc4 {100195} bxc4 {99431909} 37. Rd2 {100417} Qb3
{98947391} 38. Qg3 {100116} Bxa3 {98520214} 39. Qe5 {100548} Qb1+
{98755474} 40. Kh2 {100306} Qb8 {99905199} 41. Qxb8 {100023} Rxb8
{97895665} 42. g3 {100134} Bf8 {99558649} 43. Kg2 {100122} Rb3 {99072889}
44. Be5 {100211} Bg7 {98887411} 45. Rd8+ {100224} Kh7 {99551243} 46. Bxg7
{100538} Kxg7 {99233217} 47. Rc8 {100484} Rc3 {100000000} 48. h4 {100185}
Kf6 {98711375} 49. Rc5 {100148} e5 {93655522} 50. f4 {100525} exf4
{99856764} 51. gxf4 {100004} Kg7 {99744883} 52. e5 {100519} Rc1 {99784667}
53. Rc7 {101055} Kf8 {99927510} 54. h5 {100431} gxh5 {100000000} 55. f5
{100345} f6 {99647689} 56. Rc8+ {100358} Ke7 {98682340} 57. Rc7+ {100153}
Kd8 {99170773} 58. exf6 {100128} Ke8 {99458518} 59. f7+ {100329} Kf8
{95570653} 60. Rc6 {100303} c3 {99204990} 61. Rxh6 {100459} c2 {98194410}
62. Rc6 {100128} Kxf7 {96949815} 63. Kh2 {100094} Kg7 {97630454} 64. Rg6+
{100095} Kf8 {98796560} 65. Rc6 {100023} Kg7 {93370846} 66. Kg2 {100013} h4
{100000000} 67. Kh2 {100326} Kf7 {97965296} 68. Kg2 {100329} Kg7 {99052788}
69. Kh2 {100405} Kf7 {99225154} 70. Kg2 {100295} Ke7 {98113412} 71. Kh2
{100871} Kd7 {98002348} 72. Rc4 {100099} Kd8 {99630183} 73. f6 {100255} Kd7
{97717680} 74. f7 {100580} Ke7 {99047257} 75. f8=Q+ {100250} Kxf8
{98470075} 76. Rc7 {100558} Kg8 {98237202} 77. Rc5 {100057} Kh7 {100000000}
78. Rc6 {100144} Kg7 {98113899} 79. Rc8 {100911} Kf6 {99768781} 80. Rc7
{100895} Ke6 {100000000} 81. Rc6+ {100183} Kf7 {100000000} 82. Kg2 {100164}
Kf8 {99926653} 83. Rc7 {100270} Ke8 {100000000} 84. Rc8+ {100057} Kd7
{100000000} 85. Rc4 {100199} Ke6 {99689730} 86. Rc7 {100052} Kd6
{100000000} 87. Rc4 {100101} Ke6 {100000000} 88. Rc7 {100239} Kd6
{99953385} 89. Rc3 {100132} Ke5 {99461544} 90. Rc7 {100068} Kd5 {99772192}
91. Kh2 {100474} Kd4 {100000000} 92. Rc8 {100017} h3 {99661317} 93. Rd8+
{100092} Ke5 {100000000} 94. Rc8 {100248} Kf4 {99756585} 95. Rc7 {100241}
Ke3 {99397830} 96. Re7+ {100415} Kf3 {100000000} 97. Rf7+ {100150} Ke4
{99697634} 98. Rc7 {100708} Ke3 {100000000} 99. Re7+ {100081} Kf2
{99872765} 100. Rc7 {100054} Ke1 {100000000} 101. Re7+ {100109} Kf2
{100000000} 102. Rc7 {100283} Ke2 {99503709} 103. Rc8 {100356} Ke3
{99932902} 104. Re8+ {100163} Kf3 {99508413} 105. Rf8+ {100075} Ke3
{99431461} 106. Re8+ {100406} Kd2 {99153297} 107. Rd8+ {100103} Kc3
{98036790} 108. Rc8+ {100518} Kd4 {99444153} 109. Rd8+ {100116} Kc3
{100000000} 110. Rc8+ {100379} Kb3 {98609434} 111. Rb8+ {100341} Ka4
{99921375} 112. Rc8 {100285} Kb4 {99958865} 113. Rb8+ {100016} Ka5
{100000000} 114. Rc8 {100231} Kb4 {100000000} 115. Rb8+ {100237} Ka4
{100000000} 116. Ra8+ {100667} Kb5 {99980763} 117. Rb8+ {100139} Ka5
{100000000} 118. Ra8+ {100044} Kb5 {99647838} 119. Rb8+ {100128} Kc6
{99954061} 120. Rc8+ {100217} Kd7 {100000000} 121. Rc5 {100202} Rf1
{99823159} 122. Rxc2 {100011} 1/2-1/2
[/pgn]

[dangi12012]

4 cores for every engine
1024 mbytes hash for every engine.
stockfish 100k nodes per move
Dragon 4000k or 40000k nodes per move
KomodoDragon 4M-Stockfish100k 40-10(+31 =18 -1)
KomodoDragon 40M-Stockfish100k 35-5(+30 =10)

I wonder if there is an engine that can get 100% or close to 100% against Stockfish100k in these conditions
Maybe changing the setting of KomdoDragon can help.
I used the default parameters and did not try to increase contempt.

Why not compare the computational efficiency of both programs by calling __rdtscp on each thread?
You would need to extend the uci interface but that way you could see how many instruction cycles are needed to achieve a certain elo.
I would expect that number to go down with time.

[Chessqueen]

Dragon3Contempt50 100M nodes per move-Stockfish100k nodes per move +7 =3 -0

So it is NOT that Stockfish is better than most top programs because it calculates much much faster as it is proven here, so tell us what it is that makes Stockfish soo special ? Also what would be the time handicap that Stockfish can give Komodo Dragon 3 in order for Komodo Dragon to be equal to Stockfish like three times as much or 3 minutes to 1 minute per game or 3 seconds per move to Komodo Dragon to 1 sec per move to Stockfish:?: Can you test that as well

[Uri Blass]

Dragon3Contempt50 100M nodes per move-Stockfish100k nodes per move +7 =3 -0

So it is NOT that Stockfish is better than most top programs because it calculates much much faster as it is proven here, so tell us what it is that makes Stockfish soo special ? Also what would be the time handicap that Stockfish can give Komodo Dragon 3 in order for Komodo Dragon to be equal to Stockfish like three times as much or 3 minutes to 1 minute per game or 3 seconds per move to Komodo Dragon to 1 sec per move to Stockfish:?: Can you test that as well

My guess is that the time handicap is bigger when the time control is longer.

[Chessqueen]

Dragon3Contempt50 100M nodes per move-Stockfish100k nodes per move +7 =3 -0

So it is NOT that Stockfish is better than most top programs because it calculates much much faster as it is proven here, so tell us what it is that makes Stockfish soo special ? Also what would be the time handicap that Stockfish can give Komodo Dragon 3 in order for Komodo Dragon to be equal to Stockfish like three times as much or 3 minutes to 1 minute per game or 3 seconds per move to Komodo Dragon to 1 sec per move to Stockfish:?: Can you test that as well

My guess is that the time handicap is bigger when the time control is longer.

What GUI can I use to set different time control ?

[Uri Blass]

You can use banksia gui
Unfortunately I guess that you are going to get many draws at long time control.

For the same time control 10+0.1 with 2 cores I got 14-5 and 81 draws.
I guess that at some time control you are going to get 100 draws so it is hard to test time advantage that you need for 50%

I used contempt=0 for dragon in this test

[Uri Blass]

stockfish at 5+0.05 already lose against dragon 10+0.1

I got 7-2 and 41 draws after 50 games.

I guess at longer time control it is not going to be the case.