Slightly hairy Blitz game from nightmare.nl chess server

[chrisw]

Well, this was fun to watch, five minute blitz, sacs, material imbalance ....

[pgn][Event "ICS rated blitz match"]
[Site "nightmare-chess.nl"]
[Date "2020.09.26"]
[Round "-"]
[White "CoronaVirusChess"]
[Black "rpiRofChade"]
[Result "*"]
[WhiteElo "2705"]
[BlackElo "2843"]
[TimeControl "300+1"]
[Annotator "16. -0.29 14... +0.37"]

1. d4 {+0.40/1} Nf6 2. c4 {+0.73/1} e6 3. Nf3 {+0.99/1} d5 4. Nc3 {+0.92/1}
c6 5. e3 {+0.68/1} Nbd7 6. Bd3 {+0.99/1} dxc4 7. Bxc4 {+1.00/1} b5 8. Bd3
{+1.00/1} a6 9. O-O {+1.00/1} c5 10. e4 {+0.69/1} cxd4 11. Nxd4 {+1.00/1}
Bb7 12. a3 {+0.24/1} Nc5 13. e5 {+0.94/1} Nd5 14. Ne4 {+0.69/1} Nxd3
{+0.37/24} 15. Qxd3 {+1.00/1} Nb6 {+0.25/23} 16. Re1 {-0.29/33 15} Qd5 17.
Bg5 {+0.60/24} Na4 {+0.49/22} 18. Qd2 {+0.00/36 14} h6 {+0.34/23} 19. Bh4
{+0.31/35 7} Bc5 {+0.41/24} 20. Rad1 {+0.34/34 8} O-O {+0.40/23} 21. Qf4
{+0.00/40 12} f5 {+0.62/21} 22. Nxe6 {+0.00/42 6} Qxe6 {+0.00/28} 23. Nf6+
{+0.00/44 6} Kh8 {+0.00/24} 24. Qg3 {+0.00/43 6} Qc6 {+0.00/24} 25. Rd7
{+0.47/38 12} Qxg2+ {+0.00/26} 26. Qxg2 {+0.36/24 0.1} Bxg2 {+0.00/25} 27.
Kxg2 {+0.50/37 6} Nb6 {+0.00/25} 28. b4 {+0.49/37 7} Rxf6 {+0.00/25} 29.
Rxg7 {+0.21/37 11} Rc6 {+0.00/28} 30. Bf6 {+0.40/37 10} Bd4 {+0.00/29} 31.
Rd1 {+0.44/34 10} Rxf6 {+0.00/24} 32. exf6 {+0.44/36 10} Bb2 {-0.23/24} 33.
Rf7 {+0.63/36 5} Kg8 {-0.07/23} 34. Rc7 {+0.58/38 4} Nc4 {-0.28/22} 35. a4
{+0.51/36 5} Bxf6 {+0.00/24} 36. Rd5 {+0.55/35 4} Ne5 {+0.00/27} 37. axb5
{+0.83/34 9} axb5 {+0.00/24} 38. Rxb5 {+0.68/34 8} Ng6 {+0.00/25} 39. Ra5
{+0.35/32 8} Rd8 {+0.02/23} 40. Raa7 {+0.53/35 4} Be5 {+0.00/23} 41. Rd7
{+0.64/34 4} Rc8 {+0.00/25} 42. Ra3 {+0.72/33 4} Nf4+ {+0.00/26} 43. Kf3
{+0.89/28 7} Ng6 {+0.00/27} 44. Ra6 {+0.81/31 7} Nf4 {+0.00/28} 45. Ke3
{+0.81/33 3} Kf8 {+0.00/25} 46. Raa7 {+0.67/36 3} Re8 {-0.14/22} 47. Rab7
{+1.01/30 6} Ne6 {-0.30/24} 48. Kd3 {+0.68/29 6} Bxh2 {+0.00/20} 49. Rf7+
{+0.77/28 5} Kg8 50. Rxf5 {+0.70/30 3} h5 {+0.00/23} 51. Kc4 {+0.16/31 2.9}
h4 {+0.00/26} 52. Rf6 {+0.00/33 2.8} Nf4 {+0.00/22} 53. Rh6 {+0.42/29 5} h3
{+0.00/24} 54. Rhh7 {+0.31/33 2.7} Rc8+ {+0.00/23} 55. Kb3 {+0.29/37 2.6}
Ne6 {+0.00/25} 56. Rxh3 {+0.17/38 2.7} Nd4+ {+0.00/24} 57. Ka3
{+0.30/37 2.5} Be5 {-0.20/25} 58. Rh5 {+0.60/30 5} Bd6 {+0.00/25} 59. Ka2
{+0.76/30 4} Rc2+ {-0.05/25} 60. Kb1 {+0.80/36 2.3} Rc4 {+0.00/23} 61. Rg5+
{+0.90/30 5} Kf8 {+0.00/25} 62. Rd5 {+0.28/32 2.7} Be7 {+0.00/24} 63. b5
{+0.30/34 2.2} Bf6 {-0.23/25} 64. b6 {+0.38/36 2.1} Ke8 {+0.00/25} 65. Ka2
{+0.37/35 2.1} Ra4+ {+0.00/24} 66. Kb2 {+0.41/38 2.1} Ne6+ {+0.00/24} 67.
Kb3 {+0.22/39 4} Ra1 {+0.00/24} 68. Rb8+ {+0.27/28 2.2} Ke7 {+0.00/27} 69.
Kc2 {+0.00/33 2.8} Nd4+ {+0.00/22} 70. Kd3 {+0.00/36 1.9} Nc6 {+0.00/23}
71. Rb7+ {+0.00/36 3} Ke6 {+0.00/25} 72. Rc5 {+0.00/32 1.8} Ne5+ {+0.00/25}
73. Kc2 {+0.00/35 1.8} Ra2+ {+0.00/26} 74. Kb3 {+0.00/39 1.8} Rxf2
{+0.00/28} 75. Rbc7 {+0.00/40 2.5} Nd3 {+0.00/28} 76. R5c6+ {+0.71/37 3}
Kd5 {+0.00/35} 77. Rc2 {+0.50/39 3} Rxc2 {+0.00/38} 78. Rxc2 {+0.00/52 1.6}
Nc5+ {+0.00/45} 79. Kb4 {+0.00/46 3} Kc6 {+0.00/46} 80. b7 {+1.71/61 1.8}
Kxb7 {+0.00/97} 81. Kxc5 {+1.71/61 1.6} Bd8 {+0.00/127} 82. Kb5
{+1.71/58 2.5} Bh4 {+0.00/127} 83. Rh2 {+1.71/56 2.2} Bg5 {+0.00/127} 84.
Rh7+ {+1.71/57 1.5} Kc8 {+0.00/127} 85. Kc6 {+1.71/63 2.7} Kb8 {+0.00/127}
86. Rb7+ {+2.21/63 2.1} Ka8 {+0.00/127} 87. Rg7 {+2.36/46 1.4} Bf6
{+0.00/127} 88. Rg6 {+2.51/41 2.9} Bc3 {+0.00/127}
*
[/pgn]

[Joost Buijs]

Indeed a very nice game to watch, full of tactics.

I stopped playing OTB chess 15 years ago, 5 min. blitz was always one of my favorites, there were days that I played more than 60 blitz games in a row. It was always fun to create unbalanced positions with a lot of tactics to make it as difficult as possible for the opponent to find the right move.

Also amazing to see how strong rofChade is on a tiny RPI.

[chrisw]

Indeed a very nice game to watch, full of tactics.

I stopped playing OTB chess 15 years ago, 5 min. blitz was always one of my favorites, there were days that I played more than 60 blitz games in a row. It was always fun to create unbalanced positions with a lot of tactics to make it as difficult as possible for the opponent to find the right move.

Also amazing to see how strong rofChade is on a tiny RPI.

yup, strong engine. is the rpi also NNUE?

[Ronald]

A nice game indeed!

I think that this hairy game was played with the regular rofChade version, but I'm not totally sure, because I had to fight some battles with xboard and directories etc. The version that is playing right now is the nnue version with the "latest" net from Sergio Vieri. On the Pi the speed loss of the net is bigger than on my Threadripper, so the difference looks a little bit less. The gain on my Threadripper seems immense, around 200 elo. A testmatch against Stockfish 11 (8000 games 60-0.6) resulted in minus 14 elo difference. Like I said earlier, after 70 years of experience with traditional evaluation functions, it looks like we still don't really understand what such a function should look like...

[mvanthoor]

22. Nd4xe6 !!
23. Ne4f6 !!

I think that's a beautiful set of moves that resolves a very hairy position indeed. The black bishop/queen battery is very powerful, and the only reason white isn't mated is the knight on e4. f7f5 first looked like the death blow; either lose a piece, or be checkmated. These two moves actually put black in danger of being checkmated.

I would never have found those over the board.

[chrisw]

22. Nd4xe6 !!
23. Ne4f6 !!

I think that's a beautiful set of moves that resolves a very hairy position indeed. The black bishop/queen battery is very powerful, and the only reason white isn't mated is the knight on e4. f7f5 first looked like the death blow; either lose a piece, or be checkmated. These two moves actually put black in danger of being checkmated.

I would never have found those over the board.

Yup, I was watching in real time, thinking WTF is going on, pieces hanging everywhere. I think OTB blitz like that is really tough to handle.

[chrisw]

A nice game indeed!

I think that this hairy game was played with the regular rofChade version, but I'm not totally sure, because I had to fight some battles with xboard and directories etc. The version that is playing right now is the nnue version with the "latest" net from Sergio Vieri. On the Pi the speed loss of the net is bigger than on my Threadripper, so the difference looks a little bit less. The gain on my Threadripper seems immense, around 200 elo. A testmatch against Stockfish 11 (8000 games 60-0.6) resulted in minus 14 elo difference. Like I said earlier, after 70 years of experience with traditional evaluation functions, it looks like we still don't really understand what such a function should look like...

Rpirofchade is pretty impressive, seems to clock nodes at 1 or 2 Mnps, and it’s depth indicator is showing relatively low (well when I was watching) but the strength is very high. Which is why I wondered if it wasn’t an NNUE version. Agree with you on eval functions.

[mvanthoor]

"Rpirofchade"

Fun fact... when I first saw this engine name I read it as "Ripoffchade" ...

You know, the same way you can still read a text at almost full speed with missing or swapped letters because your brain makes up the correct words anyway. Didn't seem to work in this case

[chrisw]

This one was embarrassing. rpiRofschade at only 0.8 Mnps (I assume NNUE version), and mine got tied up in knots. If anything is going to convince me NNUE, this is it. Just awful. Evaluation is King.

[pgn][Event "ICS rated blitz match"]
[Site "nightmare-chess.nl"]
[Date "2020.09.27"]
[Round "-"]
[White "rpiRofChade"]
[Black "CoronaVirusChess"]
[Result "*"]
[WhiteElo "2845"]
[BlackElo "2701"]
[TimeControl "300+1"]
[Annotator "11. +0.43 8... -1.21"]

1. e4 e6 {+0.11/1} 2. d4 a6 {+0.22/1} 3. Bd3 b5 {+0.03/1} 4. Nf3 Bb7
{+0.81/1} 5. c3 c5 {+1.00/1} 6. O-O Nf6 {+0.70/1} 7. Re1 Be7 {+1.00/1} 8.
e5 Nd5 {-1.21/32 15} 9. dxc5 Bxc5 {-1.24/31 9} 10. a4 Qb6 {-1.11/32 14} 11.
Qe2 {+0.43/19} b4 {-1.23/31 8} 12. c4 {+0.40/17} Ne7 {-0.98/31 13} 13. Nbd2
{+0.49/20} a5 {-1.16/32 7} 14. Be4 {+0.80/17 15} Ra7 {-1.48/33 6} 15. Bxb7
{+1.12/19} Rxb7 {-1.49/36 6} 16. b3 {+1.23/18 14} d5 {-1.79/34 6} 17. exd6
{+1.32/19 13} Bxd6 {-1.79/34 11} 18. Ne4 {+1.22/21 13} O-O {-1.44/33 7} 19.
Be3 {+1.14/21 9} Qc6 {-1.50/31 10} 20. Bg5 {+1.29/22 12} Qc7 {-1.65/32 5}
21. Rad1 {+1.47/19 6} Bf4 {-1.72/35 5} 22. Bxf4 {+1.61/18 7} Qxf4
{-1.99/34 10} 23. Nc5 {+1.67/25 6} Ra7 {-1.81/44 5} 24. Nd4 {+1.54/21 11}
Qh6 {-1.84/36 5} 25. Nb5 {+1.76/20 5} Ra8 {-1.99/41 4} 26. Nd6 {+1.50/19 6}
Qf6 {-1.57/37 9} 27. Qe4 {+1.49/19 5} Nec6 {-1.85/32 8} 28. Nb5
{+1.50/18 5} Qg6 {-1.70/34 4} 29. Qe3 {+1.66/20 5} h5 {-2.20/31 8} 30. Rd6
{+2.00/20 9} h4 {-2.33/33 4} 31. h3 {+2.09/20 4} Qf6 {-2.54/32 4} 32. Ne4
{+2.33/19 9} Qg6 {-2.65/36 4} 33. Qc5 {+2.17/20 4} Ra6 {-2.90/38 4} 34. f4
{+2.43/19 8} Nd8 {-3.29/34 4} 35. Rdd1 {+2.61/20 4} Ndc6 {-2.89/32 7} 36.
Qf2 {+2.57/19 4} Ne7 {-2.76/30 4} 37. Qxh4 {+2.56/22 4} Nbc6 {-2.89/35 3}
38. Rd7 {+2.84/18 8} Nf5 {-2.64/34 6} 39. Qg5 {+2.74/18 4} Qxg5
{-2.75/33 3} 40. Nxg5 {+2.91/18 7} Raa8 {-2.72/31 3} 41. Kf2 {+3.00/18 4}
Rad8 {-3.08/33 6} 42. Red1 {+2.80/18 6} Nfe7 {-2.99/35 3} 43. g4
{+3.23/20 6} Rxd7 {-3.41/37 2.9} 44. Rxd7 {+3.71/21 6} Rd8 {-3.53/38 2.7}
45. Rxd8+ {+3.78/18 6} Nxd8 {-3.44/37 1.5} 46. Nd6 {+4.13/19 6} Nec6
{-2.84/35 5} 47. Ke3 {+4.38/18 2.8} f6 {-3.41/37 2.8} 48. Nf3
{+4.63/19 2.8} Kf8 {-3.45/36 2.8} 49. Kd3 {+4.78/20 6} Ne7 {-3.82/36 5} 50.
Ke4 {+4.86/21 2.7} Nec6 {-3.84/35 2.4} 51. c5 {+5.07/19 2.6} Ke7
{-4.00/35 2.5} 52. h4 {+5.51/20 2.5} g6 {-4.25/30 5} 53. f5 {+6.42/19 2.7}
exf5+ {-5.34/30 5} 54. gxf5 {+7.15/22 5} g5 {-6.14/33 3} 55. hxg5
{+7.59/21 2.9} fxg5 {-6.33/35 2.8} 56. Nxg5 {+7.40/21 5} Nb8 {-6.55/35 2.8}
57. Kd5 {+7.82/20 2.2} Na6 {-6.88/35 2.5} 58. Ne6 {+8.44/21 4} Nxe6
{-7.67/40 4} 59. fxe6 {+8.23/24 2.4} Nc7+ {-9.00/37 3} 60. Kc4
{+8.28/23 2.1} Kxe6 {-9.43/38 4} 61. Nb7 {+8.50/23 2.2} Ke7 {-9.91/40 2.2}
62. Nxa5 {+8.64/21 4} Kd7 {-10.36/39 2.2} 63. Kxb4 {+8.64/21 2.1} Nd5+
{-11.69/35 2.6} 64. Kc4 {+8.93/21 4}
*
[/pgn]

[mvanthoor]

I wonder, if one would take a very basic engine, and then add NNUE to it, what the result would be. Will a simple engine also gain 200 ELO (just like it will gain about 80-100 if you give it 4 cores instead of one), or will it gain nothing, because NNUE only works well if your normal evaluation is relatively decent? (I don't know; I haven't looked into NNUE yet. Probably won't for a very long time.)