Super bullet chess: LC0 network 31740 vs Stockfish 10 (~3500 elo) drawkiller tournament pgn

[Hai]

Super bullet chess: LC0 network 31740 vs Stockfish 10 (~3500 elo)

Drawkiller tournament pgn
After 40 games (still running)
Ponder on
LC0 with RTX 2080 Ti 32 kn/s
Stockfish 10 with 4 cores @ 4.2 ghz
Time control: 1 second per move.

Result:
LC0 18.5 points
Stockfish 21.5 points
+12 =13 -15
Winning percentage = 46.25% compare to http://talkchess.com/forum3/viewtopic.php?f=6&t=69120

= LC0 is 26 elo weaker than Stockfish 10.

[Hai]

After 105 games:

Result:
LC0 48.5
Stockfish 56.5
+30 =37 -38
Winning percentage = 46.19%

= LC0 is 27 elo weaker than Stockfish 10.

Match is still running.

[Hai]

After 287 games:

Result:
LC0 137
Stockfish 150
+85 =104 -98
Winning percentage = 47.74%

= LC0 is 16 elo weaker than Stockfish 10.
Feel free to compare drawkiller tournament pgn with the 20 possible opening moves.
http://talkchess.com/forum3/viewtopic.php?f=6&t=69120

[lucasart]

After 287 games:

Result:
LC0 137
Stockfish 150
+85 =104 -98
Winning percentage = 47.74%

= LC0 is 16 elo weaker than Stockfish 10.
Feel free to compare drawkiller tournament pgn with the 20 possible opening moves.
http://talkchess.com/forum3/viewtopic.php?f=6&t=69120

Is this the LC0 fake news network ? LC0 is nowhere near SF…

[mwyoung]

Theory and practice sometimes clash. And when this happens. Theory loses every time.

In practice Lc0 is closing in on Stockfish 10. Not in just my testing, but others. Practice wins again.

[Javier Ros]

Super bullet chess: LC0 network 31740 vs Stockfish 10 (~3500 elo)

Drawkiller tournament pgn
After 40 games (still running)
Ponder on
LC0 with RTX 2080 Ti 32 kn/s
Stockfish 10 with 4 cores @ 4.2 ghz
Time control: 1 second per move.

Result:
LC0 18.5 points
Stockfish 21.5 points
+12 =13 -15
Winning percentage = 46.25% compare to http://talkchess.com/forum3/viewtopic.php?f=6&t=69120

= LC0 is 26 elo weaker than Stockfish 10.

Can you publish the games?
It would be very interesting.

[pohl4711]

After 287 games:

Result:
LC0 137
Stockfish 150
+85 =104 -98
Winning percentage = 47.74%

= LC0 is 16 elo weaker than Stockfish 10.
Feel free to compare drawkiller tournament pgn with the 20 possible opening moves.
http://talkchess.com/forum3/viewtopic.php?f=6&t=69120

Nice to see, how good my Drawkiller openings work. After 287 games using Drawkiller, the draw-rate is 36%
40 games with the 20 possible opening moves (+7 =24 -9), the draw-rate is 60% and these 20 possible openingmoves contain strange moves like 1.a4 or 1.h4, which should lower the draw-rate,too...(with "classical, normal" openings, the draw-rate would be even higher than 60%))

Stefan (SPCC)

[Hai]

Super bullet chess: LC0 network 31740 vs Stockfish 10 (~3500 elo)

Drawkiller tournament pgn
After 40 games (still running)
Ponder on
LC0 with RTX 2080 Ti 32 kn/s
Stockfish 10 with 4 cores @ 4.2 ghz
Time control: 1 second per move.

Result:
LC0 18.5 points
Stockfish 21.5 points
+12 =13 -15
Winning percentage = 46.25% compare to http://talkchess.com/forum3/viewtopic.php?f=6&t=69120

= LC0 is 26 elo weaker than Stockfish 10.

Can you publish the games?
It would be very interesting.

http://www.mediafire.com/file/uf7q83ej3 ... s.pgn/file

[Javier Ros]

Super bullet chess: LC0 network 31740 vs Stockfish 10 (~3500 elo)

Drawkiller tournament pgn
After 40 games (still running)
Ponder on
LC0 with RTX 2080 Ti 32 kn/s
Stockfish 10 with 4 cores @ 4.2 ghz
Time control: 1 second per move.

Result:
LC0 18.5 points
Stockfish 21.5 points
+12 =13 -15
Winning percentage = 46.25% compare to http://talkchess.com/forum3/viewtopic.php?f=6&t=69120

= LC0 is 26 elo weaker than Stockfish 10.

Can you publish the games?
It would be very interesting.

http://www.mediafire.com/file/uf7q83ej3 ... s.pgn/file

Thanks.