Natural TB

mcostalba · Post by **mcostalba** » Fri Jun 03, 2016 6:45 am

jhellis3 wrote:Here is the fix applied to vanilla SF if anyone is interested:

https://github.com/jhellis3/Stockfish/tree/tb_fix_1

[d]8/2P1P3/3k4/8/8/4K3/P2p1p2/8 b - -

512MB Hash, Single Thread:

Vanilla (with TBs) @ ~200 seconds: Depth 51, score +123.xx, short line displayed

Patch @ ~200 seconds: Depth 49, score Mate in 14, full line displayed (starts returning mate scores at d31)

Vanilla (sans TBs) @ ~80 seconds: Depth 40, score Mate in 14, full line displayed (starts returning mate scores at d35)

Natural TB finds the mate in 14 in at depth 46 (same conditions as yours 512MB hash, single core). Starts returning mate scores at d30

Code: Select all

1...f2-f1Q 2.c7-c8Q Qf1-e1+ 3.Ke3-f4 Qe1xe7 
  -+  (-2.06)   Depth: 2/2   00:00:00
1...d2-d1Q 2.c7-c8N+ Kd6-d7 3.e7-e8B+ Kd7xe8 
  -+  (-6.46)   Depth: 3/5   00:00:00
1...d2-d1Q 2.c7-c8N+ Kd6-d7 3.e7-e8B+ Kd7xe8 
  -+  (-6.46)   Depth: 4/6   00:00:00
1...d2-d1Q 2.c7-c8N+ Kd6-d7 3.e7-e8B+ Kd7xe8 
  -+  (-6.46)   Depth: 4/6   00:00:00
1...d2-d1Q 2.e7-e8N+ Kd6-e7 3.Ke3xf2 Qd1-c2+ 4.Kf2-e3 Ke7xe8 
  -+  (-7.87)   Depth: 5/7   00:00:00
1...d2-d1Q 2.e7-e8N+ Kd6-e7 3.Ke3xf2 Qd1-c2+ 4.Kf2-e3 Ke7xe8 5.c7-c8N Qc2xc8 
  -+  (-9.45)   Depth: 6/8   00:00:00  10kN, tb=394
1...d2-d1Q 2.e7-e8N+ Kd6-e7 3.Ke3xf2 Qd1-c2+ 4.Kf2-e3 Ke7xe8 5.c7-c8N Qc2xc8 
  -+  (-9.45)   Depth: 7/10   00:00:00  15kN, tb=647
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d4+ 5.Kf2-f3 Qd4-d5+ 6.Kf3-e2 Qd5-c4+ 7.Ke2-e3 
  -+  (-9.37)   Depth: 8/12   00:00:00  37kN, tb=2583
1...d2-d1Q 2.c7-c8N+ Kd6-d7 3.e7-e8B+ Kd7xe8 4.Ke3xf2 Qd1-d2+ 5.Kf2-g3 Qd2-c3+ 6.Kg3-h2 Qc3-c2+ 7.Kh2-h1 Qc2-c1+ 8.Kh1-h2 
  -+  (-11.00)   Depth: 9/18   00:00:00  117kN, tb=14075
1...d2-d1Q 2.c7-c8N+ Kd6-d7 3.Nc8-b6+ Kd7xe7 4.Nb6-c8+ Ke7-e8 5.Ke3-f4 Qd1-c1+ 6.Kf4-e5 Qc1-c3+ 7.Ke5-e4 Qc3-c2+ 8.Ke4-d4 Qc2xc8 9.a2-a4 
  -+  (-38.06)   Depth: 10/22   00:00:00  314kN, tb=26724
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8R Kd7xc8 4.Ke3xf2 Qd1-d2+ 5.Kf2-f3 Qd2-d5+ 6.Kf3-f4 Qd5-f7+ 7.Kf4-e3 Qf7-e7+ 8.Ke3-f2 Qe7-a7+ 9.Kf2-e1 Qa7-a5+ 10.Ke1-d1 Qa5-a4+ 11.Kd1-c1 Qa4-f4+ 12.Kc1-b1 Qf4-b4+ 13.Kb1-c1 Qb4-c3+ 14.Kc1-b1 
  -+  (-38.72)   Depth: 11/30   00:00:00  890kN, tb=78695
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.Ke3xf2 Qd1-c2+ 4.Kf2-e3 Qc2-c3+ 5.Ke3-e2 Kd7xe8 6.a2-a4 Qc3-c2+ 7.Ke2-e3 Qc2-c5+ 8.Ke3-e2 Qc5-e5+ 9.Ke2-f2 Qe5-f4+ 10.Kf2-g2 Qf4-g4+ 11.Kg2-h2 Qg4-e2+ 12.Kh2-h3 Qe2-e6+ 13.Kh3-h2 Qe6-h6+ 14.Kh2-g1 Qh6-c1+ 15.Kg1-g2 
  -+  (-38.66)   Depth: 12/30   00:00:00  1215kN, tb=105074
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.Ke3xf2 Qd1-c2+ 4.Kf2-e3 Qc2-c3+ 5.Ke3-e2 Kd7xe8 6.c7-c8R+ Qc3xc8 7.Ke2-d2 Qc8-d8+ 8.Kd2-e2 Qd8-e7+ 9.Ke2-d2 Qe7-g5+ 10.Kd2-c2 Qg5-c5+ 11.Kc2-b2 Qc5-b4+ 12.Kb2-c2 Ke8-d7 
  -+  (-38.73)   Depth: 13/32   00:00:00  1333kN, tb=115983
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-h5 5.Ne8-f6 Qh5-h2+ 6.Kf2-e3 Qh2-h6+ 7.Ke3-d3 Qh6xf6 8.Kd3-c2 Qf6-g6+ 9.Kc2-c3 Qg6-g7+ 10.Kc3-b3 Qg7-f7+ 11.Kb3-a3 Qf7-f3+ 12.Ka3-a4 Qf3-a8+ 13.Ka4-b3 Qa8-d5+ 14.Kb3-a3 Qd5-d3+ 15.Ka3-b4 
  -+  (-38.75)   Depth: 14/32   00:00:00  1420kN, tb=122756
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-h5 5.Ne8-f6 Qh5-h2+ 6.Kf2-e3 Qh2-h6+ 7.Ke3-d3 Qh6xf6 8.Kd3-c2 Qf6-g6+ 9.Kc2-c3 Qg6-g7+ 10.Kc3-b3 Qg7-f7+ 11.Kb3-a3 Qf7-f3+ 12.Ka3-a4 Qf3-d1+ 13.Ka4-a3 Qd1-c1+ 14.Ka3-a4 Qc1-c4+ 15.Ka4-a3 
  -+  (-38.75)   Depth: 15/32   00:00:00  1474kN, tb=124184
1...d2-d1Q 2.c7-c8N+ Kd6-d7 3.Ke3xf2 Qd1-c2+ 4.Kf2-e3 Qc2xc8 5.e7-e8R Qc8xe8+ 6.Ke3-d4 Qe8-h8+ 7.Kd4-d3 Qh8-g8 8.Kd3-d4 Qg8xa2 9.Kd4-e3 Qa2-d5 10.Ke3-e2 Kd7-d6 11.Ke2-e3 
  -+  (-48.95)   Depth: 16/34   00:00:00  1939kN, tb=145982
1...d2-d1Q 2.c7-c8N+ Kd6-d7 3.Ke3xf2 Qd1-c2+ 4.Kf2-e3 Qc2xc8 5.e7-e8R Qc8xe8+ 6.Ke3-d4 Qe8-h8+ 7.Kd4-d3 Qh8-g8 8.Kd3-d4 Qg8xa2 9.Kd4-e3 Qa2-d5 10.Ke3-e2 Kd7-d6 11.Ke2-e3 Qd5-e6+ 12.Ke3-d3 
  -+  (-48.95)   Depth: 17/34   00:00:01  2122kN, tb=146850
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Qa3-a2+ 11.Kd2-e3 Kd7-d6 12.Ke3-d3 Qa2-f2 
  -+  (-49.03)   Depth: 18/34   00:00:01  2526kN, tb=147843
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Qa3-a2+ 11.Kd2-e3 Kd7-d6 12.Ke3-d3 Qa2-f2 13.Kd3-c3 
  -+  (-49.03)   Depth: 19/34   00:00:01  2895kN, tb=150352
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Qa3-a2+ 11.Kd2-e3 Kd7-d6 12.Ke3-d3 Qa2-f2 13.Kd3-e4 Qf2-e2+ 14.Ke4-d4 Qe2-d2+ 15.Kd4-e4 Qd2-c3 
  -+  (-49.03)   Depth: 20/34   00:00:01  3464kN, tb=153053
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Qa3-a2+ 11.Kd2-e3 Kd7-d6 12.Ke3-d3 Qa2-f2 13.Kd3-e4 Qf2-e2+ 14.Ke4-d4 Qe2-d2+ 15.Kd4-e4 
  -+  (-49.03)   Depth: 21/34   00:00:01  4156kN, tb=157176
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Qa3-a2+ 11.Kd2-e3 Kd7-d6 12.Ke3-d3 Qa2-f2 13.Kd3-e4 Qf2-e2+ 14.Ke4-d4 Qe2-d2+ 15.Kd4-e4 Qd2-e1+ 16.Ke4-d3 Qe1-d1+ 17.Kd3-c3 
  -+  (-49.03)   Depth: 22/34   00:00:02  5170kN, tb=163209
1...d2-d1Q 
  -+  (-49.10)   Depth: 23/34   00:00:03  7126kN, tb=165656
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Kd7-d6 11.Kd2-e2 Qa3-a5 12.Ke2-d1 Qa5-a4+ 13.Kd1-c1 Qa4-c4+ 14.Kc1-d2 Qc4-b4+ 15.Kd2-e3 Kd6-d5 16.Ke3-f2 
  -+  (-49.11)   Depth: 23/34   00:00:03  7842kN, tb=169727
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Kd7-d6 11.Kd2-e2 Qa3-a5 12.Ke2-d1 Qa5-a4+ 13.Kd1-d2 Qa4-b4+ 14.Kd2-e3 Kd6-d5 15.Ke3-f2 Qb4-f8+ 16.Kf2-e3 Qf8-a3+ 17.Ke3-e2 
  -+  (-49.11)   Depth: 24/34   00:00:04  9769kN, tb=178700
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8B+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Kf2-f3 Qd8xe8 6.a2-a3 Qe8-f8+ 7.Kf3-e2 Qf8-e7+ 8.Ke2-f3 Qe7xa3+ 9.Kf3-e2 Kc8-d7 10.Ke2-d2 Kd7-d6 11.Kd2-e2 Qa3-a5 12.Ke2-e3 Qa5-a2 13.Ke3-d3 Qa2-b2 14.Kd3-e3 Qb2-b3+ 15.Ke3-e4 Qb3-b4+ 16.Ke4-e3 Kd6-d5 
  -+  (-49.11)   Depth: 25/34   00:00:05  12371kN, tb=192467
1...d2-d1Q 
  -+  (-49.18)   Depth: 26/34   00:00:06  15801kN, tb=196207
1...d2-d1Q 
  -+  (-49.25)   Depth: 26/34   00:00:07  17457kN, tb=198581
1...d2-d1Q 
  -+  (-49.35)   Depth: 26/34   00:00:07  18165kN, tb=200848
1...d2-d1Q 
  -+  (-49.50)   Depth: 26/34   00:00:07  18824kN, tb=201493
1...d2-d1Q 
  -+  (-49.70)   Depth: 26/34   00:00:07  19204kN, tb=201721
1...d2-d1Q 
  -+  (-49.97)   Depth: 26/34   00:00:07  19821kN, tb=202620
1...d2-d1Q 
  -+  (-50.33)   Depth: 26/34   00:00:08  19998kN, tb=202755
1...d2-d1Q 
  -+  (-50.79)   Depth: 26/34   00:00:08  20408kN, tb=203372
1...d2-d1Q 
  -+  (-51.39)   Depth: 26/34   00:00:08  20551kN, tb=203413
1...d2-d1Q 
  -+  (-52.16)   Depth: 26/34   00:00:08  20910kN, tb=204001
1...d2-d1Q 
  -+  (-53.14)   Depth: 26/36   00:00:08  21599kN, tb=206535
1...d2-d1Q 
  -+  (-54.38)   Depth: 26/36   00:00:09  22260kN, tb=207538
1...d2-d1Q 
  -+  (-55.95)   Depth: 26/36   00:00:09  22828kN, tb=208763
1...d2-d1Q 
  -+  (-57.93)   Depth: 26/36   00:00:09  23398kN, tb=209027
1...d2-d1Q 2.e7-e8N+ 
  -+  (-54.75)   Depth: 26/36   00:00:09  24096kN, tb=226630
1...d2-d1Q 
  -+  (-58.36)   Depth: 26/36   00:00:09  24213kN, tb=226651
1...d2-d1Q 
  -+  (-62.29)   Depth: 26/36   00:00:10  24390kN, tb=226665
1...d2-d1Q 
  -+  (-67.22)   Depth: 26/36   00:00:10  24553kN, tb=226754
1...d2-d1Q 
  -+  (-73.41)   Depth: 26/36   00:00:10  24738kN, tb=227363
1...d2-d1Q 
  -+  (-81.16)   Depth: 26/36   00:00:10  24806kN, tb=227365
1...d2-d1Q 
  -+  (-90.87)   Depth: 26/36   00:00:10  24859kN, tb=227365
1...d2-d1Q 
  -+  (-103.03)   Depth: 26/36   00:00:10  24913kN, tb=227377
1...d2-d1Q 
  -+  (-118.24)   Depth: 26/36   00:00:10  24963kN, tb=227377
1...d2-d1Q 2.e7-e8N+ 
  -+  (-69.52)   Depth: 26/36   00:00:10  25226kN, tb=229197
1...d2-d1Q 
  -+  (-114.80)   Depth: 26/36   00:00:10  25253kN, tb=229231
1...d2-d1Q 2.e7-e8N+ 
  -+  (-69.52)   Depth: 26/36   00:00:10  25541kN, tb=229497
1...d2-d1Q 
  -+  (-102.14)   Depth: 26/36   00:00:10  25546kN, tb=229498
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Ne8-g7 Qd8-f8+ 6.Kf2-e2 Qf8xg7 7.a2-a4 Qg7-b2+ 8.Ke2-f1 Qb2-d2 9.a4-a5 Qd2xa5 10.Kf1-e2 Kc8-d7 11.Ke2-e3 Qa5-a3+ 12.Ke3-f2 Kd7-e7 13.Kf2-g2 Qa3-a2+ 14.Kg2-f3 Qa2-a4 15.Kf3-g3 Ke7-d7 16.Kg3-f3 Qa4-a2 17.Kf3-g4 
  -+  (-80.26)   Depth: 26/36   00:00:10  25559kN, tb=229569
1...d2-d1Q 
  -+  (-80.33)   Depth: 27/36   00:00:10  26117kN, tb=234080
1...d2-d1Q 
  -+  (-80.40)   Depth: 27/36   00:00:10  26337kN, tb=234159
1...d2-d1Q 
  -+  (-80.50)   Depth: 27/36   00:00:10  26359kN, tb=234189
1...d2-d1Q 
  -+  (-80.85)   Depth: 27/36   00:00:10  26402kN, tb=234191
1...d2-d1Q 
  -+  (-80.85)   Depth: 27/36   00:00:10  26402kN, tb=234191
1...d2-d1Q 
  -+  (-81.12)   Depth: 27/36   00:00:10  26440kN, tb=234191
1...d2-d1Q 
  -+  (-81.48)   Depth: 27/38   00:00:10  26481kN, tb=234191
1...d2-d1Q 
  -+  (-81.94)   Depth: 27/38   00:00:11  26509kN, tb=234191
1...d2-d1Q 
  -+  (-82.55)   Depth: 27/38   00:00:11  26535kN, tb=234191
1...d2-d1Q 
  -+  (-83.31)   Depth: 27/38   00:00:11  26557kN, tb=234192
1...d2-d1Q 
  -+  (-84.29)   Depth: 27/38   00:00:11  26588kN, tb=234192
1...d2-d1Q 
  -+  (-85.53)   Depth: 27/38   00:00:11  26615kN, tb=234192
1...d2-d1Q 
  -+  (-87.10)   Depth: 27/38   00:00:11  26643kN, tb=234192
1...d2-d1Q 
  -+  (-89.08)   Depth: 27/38   00:00:11  26668kN, tb=234192
1...d2-d1Q 
  -+  (-91.57)   Depth: 27/38   00:00:11  26749kN, tb=234656
1...d2-d1Q 
  -+  (-94.70)   Depth: 27/38   00:00:11  26779kN, tb=234656
1...d2-d1Q 
  -+  (-98.64)   Depth: 27/38   00:00:11  26804kN, tb=234657
1...d2-d1Q 
  -+  (-103.57)   Depth: 27/38   00:00:11  26836kN, tb=234657
1...d2-d1Q 
  -+  (-109.76)   Depth: 27/38   00:00:11  27095kN, tb=235397
1...d2-d1Q 
  -+  (-117.51)   Depth: 27/38   00:00:11  27379kN, tb=238310
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Ne8-g7 Qd8-f8+ 6.Kf2-e2 Qf8xg7 7.a2-a4 Qg7-b2+ 8.Ke2-e1 Qb2-a2 9.a4-a5 Qa2xa5+ 10.Ke1-e2 Qa5-a2+ 11.Ke2-d3 Kc8-d7 12.Kd3-e3 Kd7-e7 13.Ke3-f3 Ke7-d6 14.Kf3-e3 Qa2-a3+ 15.Ke3-f2 Qa3-d3 16.Kf2-g1 Qd3-b1+ 17.Kg1-f2 
  -+  (-115.57)   Depth: 27/38   00:00:11  27768kN, tb=247868
1...d2-d1Q 
  -+  (-115.64)   Depth: 28/38   00:00:12  29093kN, tb=252830
1...d2-d1Q 
  -+  (-115.71)   Depth: 28/38   00:00:12  29120kN, tb=252834
1...d2-d1Q 
  -+  (-115.82)   Depth: 28/38   00:00:12  29181kN, tb=252895
1...d2-d1Q 
  -+  (-115.96)   Depth: 28/39   00:00:12  29197kN, tb=252895
1...d2-d1Q 
  -+  (-116.17)   Depth: 28/39   00:00:12  29241kN, tb=253121
1...d2-d1Q 
  -+  (-116.44)   Depth: 28/39   00:00:12  29472kN, tb=254681
1...d2-d1Q 
  -+  (-116.79)   Depth: 28/39   00:00:12  29533kN, tb=254686
1...d2-d1Q 
  -+  (-117.26)   Depth: 28/39   00:00:12  29798kN, tb=255251
1...d2-d1Q 
  -+  (-117.86)   Depth: 28/39   00:00:12  29876kN, tb=255388
1...d2-d1Q 
  -+  (-118.63)   Depth: 28/39   00:00:12  30197kN, tb=256060
1...d2-d1Q 
  -+  (-119.60)   Depth: 28/39   00:00:12  30401kN, tb=256138
1...d2-d1Q 
  -+  (-120.84)   Depth: 28/39   00:00:12  30704kN, tb=256430
1...d2-d1Q 
  -+  (-122.41)   Depth: 28/39   00:00:13  31215kN, tb=256911
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Ne8-g7 Qd8-f8+ 6.Kf2-g3 Qf8xg7+ 7.Kg3-h4 Qg7-a1 8.a2-a4 Qa1xa4+ 9.Kh4-g3 Qa4-a3+ 10.Kg3-f4 Kc8-c7 11.Kf4-e4 Kc7-d6 12.Ke4-d4 Qa3-b3 13.Kd4-e4 Qb3-c2+ 14.Ke4-f3 Qc2-d3+ 15.Kf3-g2 Kd6-c7 16.Kg2-g1 Qd3-d2 17.Kg1-f1 
  -+  (-123.13)   Depth: 29/43   00:00:14  33013kN, tb=277318
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Ne8-g7 Qd8-f8+ 6.Kf2-g3 Qf8xg7+ 7.Kg3-h4 Qg7-a1 8.a2-a4 Qa1xa4+ 9.Kh4-g3 Qa4-a3+ 10.Kg3-f4 Kc8-c7 11.Kf4-e4 Kc7-d6 12.Ke4-d4 Qa3-b3 13.Kd4-e4 Qb3-c2+ 14.Ke4-f3 Qc2-d3+ 15.Kf3-g2 Kd6-c7 16.Kg2-g1 Qd3-d2 17.Kg1-f1 
  -+  (-123.13)   Depth: 29/43   00:00:15  36428kN, tb=296265
1...d2-d1Q 
  -+  (-123.20)   Depth: 30/43   00:00:16  37414kN, tb=297882
1...d2-d1Q 
  -+  (-123.27)   Depth: 30/43   00:00:16  37909kN, tb=298092
1...d2-d1Q 
  -+  (-123.38)   Depth: 30/43   00:00:16  38321kN, tb=298153
1...d2-d1Q 
  -+  (-123.53)   Depth: 30/43   00:00:16  38684kN, tb=298308
1...d2-d1Q 
  -+  (-#39)   Depth: 30/43   00:00:17  39385kN, tb=299421
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d8 5.Ne8-g7 Qd8-f8+ 6.Kf2-g3 Qf8xg7+ 7.Kg3-h4 Qg7-a1 8.a2-a4 Qa1xa4+ 9.Kh4-g3 Kc8-d7 10.Kg3-f2 Kd7-e8 11.Kf2-e3 Qa4-a3+ 12.Ke3-d4 Qa3-a2 13.Kd4-c5 Ke8-e7 14.Kc5-b6 Qa2-d5 15.Kb6-a6 Qd5-c6+ 16.Ka6-a5 Qc6-c4 17.Ka5-b6 
  -+  (-#44)   Depth: 31/46   00:00:18  40853kN, tb=324040
1...d2-d1Q 
  -+  (-#35)   Depth: 31/46   00:00:19  42658kN, tb=327382
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d5 5.Ne8-f6 Qd5-d4+ 6.Kf2-g2 Qd4xf6 7.a2-a4 Qf6-c6+ 8.Kg2-h3 Qc6-d7+ 9.Kh3-g3 Qd7xa4 10.Kg3-f3 Qa4-a3+ 11.Kf3-e2 Qa3-g3 12.Ke2-d2 Qg3-f3 13.Kd2-c2 Qf3-e3 14.Kc2-b2 Qe3-d3 15.Kb2-c1 Kc8-c7 16.Kc1-b2 Kc7-d6 17.Kb2-a1 
  -+  (-#28)   Depth: 31/46   00:00:21  46462kN, tb=358588
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d5 5.Kf2-e3 Qd5-e5+ 6.Ke3-d2 Qe5xe8 7.Kd2-c1 Qe8-a4 8.Kc1-b2 Kc8-d7 9.Kb2-b1 Qa4-a3 10.Kb1-c2 Qa3xa2+ 11.Kc2-d3 Qa2-a3+ 12.Kd3-d2 Kd7-c7 13.Kd2-e2 Qa3-g3 14.Ke2-d2 Qg3-f3 15.Kd2-c2 Qf3-e3 16.Kc2-b2 Qe3-d3 17.Kb2-a2 
  -+  (-#28)   Depth: 32/46   00:00:24  51809kN, tb=411307
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d5 5.Kf2-e3 Qd5-e5+ 6.Ke3-d2 Qe5xe8 7.Kd2-c2 Qe8-a4+ 8.Kc2-b2 Kc8-c7 9.Kb2-b1 Qa4-a3 10.Kb1-c2 Qa3xa2+ 11.Kc2-d3 Qa2-a3+ 12.Kd3-e2 Kc7-c8 13.Ke2-d2 Qa3-f3 14.Kd2-c2 Qf3-e3 15.Kc2-b2 Qe3-d3 16.Kb2-c1 Kc8-c7 17.Kc1-b2 
  -+  (-#28)   Depth: 33/46   00:00:26  55851kN, tb=464272
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d7 5.Ne8-f6 Qd7-f5+ 6.Kf2-g2 Qf5xf6 7.Kg2-g3 Qf6-a6 8.Kg3-f3 Qa6xa2 9.Kf3-e3 Qa2-a3+ 10.Ke3-e2 Kc8-c7 11.Ke2-d2 Qa3-f3 12.Kd2-c2 Qf3-e3 13.Kc2-d1 Qe3-d3+ 14.Kd1-e1 Qd3-c2 15.Ke1-f1 Qc2-h2 16.Kf1-e1 Qh2-g2 17.Ke1-d1 
  -+  (-#24)   Depth: 34/46   00:00:28  60633kN, tb=533089
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d7 5.Ne8-f6 Qd7-f5+ 6.Kf2-g2 Qf5xf6 7.Kg2-g3 Qf6-a6 8.Kg3-f3 Qa6xa2 9.Kf3-e3 Qa2-a3+ 10.Ke3-e2 Kc8-c7 11.Ke2-d2 Qa3-f3 12.Kd2-c2 Qf3-e3 13.Kc2-d1 Qe3-d3+ 14.Kd1-e1 Qd3-c2 15.Ke1-f1 Kc7-d6 16.Kf1-e1 Qc2-g2 17.Ke1-d1 
  -+  (-#22)   Depth: 35/46   00:00:33  68483kN, tb=618159
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d7 5.Ne8-f6 Qd7-f5+ 6.Kf2-g2 Qf5xf6 7.Kg2-g3 Qf6-a6 8.Kg3-f3 Qa6xa2 9.Kf3-e3 Qa2-a3+ 10.Ke3-e2 Qa3-c3 11.Ke2-f2 Qc3-d2+ 12.Kf2-g3 Qd2-e2 13.Kg3-h3 Kc8-d7 14.Kh3-g3 Kd7-d6 15.Kg3-h3 Qe2-f3+ 16.Kh3-h2 Kd6-d5 17.Kh2-g1 
  -+  (-#21)   Depth: 36/46   00:00:36  76064kN, tb=717479
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d7 5.Ne8-f6 Qd7-f5+ 6.Kf2-g2 Qf5xf6 7.a2-a4 Qf6-c6+ 8.Kg2-f2 Qc6xa4 9.Kf2-e2 Qa4-a3 10.Ke2-d2 Qa3-f3 11.Kd2-c2 Qf3-e3 12.Kc2-b2 Kc8-d7 13.Kb2-c2 Kd7-d6 14.Kc2-b1 Qe3-d2 15.Kb1-a1 Kd6-d5 16.Ka1-b1 Kd5-c4 17.Kb1-a1 
  -+  (-#18)   Depth: 37/46   00:00:42  88101kN, tb=882964
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d7 5.Ne8-f6 Qd7-f5+ 6.Kf2-g2 Qf5xf6 7.a2-a4 Qf6-c6+ 8.Kg2-f2 Qc6xa4 9.Kf2-e2 Qa4-a3 10.Ke2-d2 Qa3-f3 11.Kd2-c2 Kc8-d7 12.Kc2-b2 Kd7-d6 13.Kb2-c2 Kd6-d5 14.Kc2-d2 Kd5-c4 15.Kd2-e1 Qf3-g2 16.Ke1-d1 Kc4-d3 17.Kd1-e1 
  -+  (-#17)   Depth: 38/46   00:00:47  98873kN, tb=982091
1...d2-d1Q 2.e7-e8N+ Kd6-d7 3.c7-c8Q+ Kd7xc8 4.Ke3xf2 Qd1-d7 5.Ne8-f6 Qd7-f5+ 6.Kf2-g2 Qf5xf6 7.a2-a4 Qf6-c6+ 8.Kg2-f2 Qc6xa4 9.Kf2-e2 Qa4-a3 10.Ke2-d2 Qa3-f3 11.Kd2-c2 Kc8-d7 12.Kc2-b2 Kd7-d6 13.Kb2-c2 Kd6-d5 14.Kc2-d2 Kd5-c4 15.Kd2-e1 Qf3-g2 16.Ke1-d1 Kc4-d3 17.Kd1-e1 
  -+  (-#17)   Depth: 39/46   00:00:53  110mN, tb=1081955
1...f2-f1Q 2.c7-c8N+ Kd6-d7 3.Ke3xd2 Qf1-a6 4.Nc8-b6+ Qa6xb6 5.e7-e8R Kd7xe8 6.Kd2-e2 Qb6-e6+ 7.Ke2-d2 Qe6xa2+ 8.Kd2-d3 Ke8-e7 9.Kd3-e3 Qa2-c2 10.Ke3-f4 Qc2-e2 11.Kf4-f5 Qe2-e3 12.Kf5-g6 Ke7-e6 13.Kg6-h5 Qe3-g3 14.Kh5-h6 Ke6-f6 15.Kh6-h5 Qg3-h3# 
  -+  (-#15)   Depth: 40/46   00:01:57  265mN, tb=1928309
1...f2-f1Q 2.c7-c8N+ Kd6-d7 3.Ke3xd2 Qf1-a6 4.Nc8-b6+ Qa6xb6 5.e7-e8R Kd7xe8 6.Kd2-e2 Qb6-e6+ 7.Ke2-d2 Qe6xa2+ 8.Kd2-d3 Ke8-e7 9.Kd3-e3 Qa2-c2 10.Ke3-f4 Qc2-e2 11.Kf4-f5 Qe2-e3 12.Kf5-g6 Ke7-e6 13.Kg6-h5 Qe3-g3 14.Kh5-h6 Ke6-f6 15.Kh6-h5 Qg3-h3# 
  -+  (-#15)   Depth: 41/46   00:02:07  289mN, tb=2059625
1...f2-f1Q 2.c7-c8N+ Kd6-d7 3.Ke3xd2 Qf1-a6 4.Nc8-b6+ Qa6xb6 5.e7-e8R Kd7xe8 6.Kd2-e2 Qb6-e6+ 7.Ke2-d2 Qe6xa2+ 8.Kd2-d3 Ke8-e7 9.Kd3-e3 Qa2-c2 10.Ke3-f4 Qc2-e2 11.Kf4-f5 Qe2-e3 12.Kf5-g6 Ke7-e6 13.Kg6-h5 Qe3-g3 14.Kh5-h6 Ke6-f6 15.Kh6-h5 Qg3-h3# 
  -+  (-#15)   Depth: 42/46   00:02:19  316mN, tb=2228498
1...f2-f1Q 2.c7-c8N+ Kd6-d7 3.Ke3xd2 Qf1-a6 4.Nc8-b6+ Qa6xb6 5.e7-e8R Kd7xe8 6.Kd2-e2 Qb6-e6+ 7.Ke2-d2 Qe6xa2+ 8.Kd2-d3 Ke8-e7 9.Kd3-e3 Qa2-c2 10.Ke3-f4 Qc2-e2 11.Kf4-f5 Qe2-e3 12.Kf5-g6 Ke7-e6 13.Kg6-h5 Qe3-g3 14.Kh5-h6 Ke6-f6 15.Kh6-h5 Qg3-h3# 
  -+  (-#15)   Depth: 43/46   00:02:35  353mN, tb=2539393
1...f2-f1Q 2.c7-c8N+ Kd6-d7 3.Ke3xd2 Qf1-a6 4.Nc8-b6+ Qa6xb6 5.e7-e8R Kd7xe8 6.Kd2-e2 Qb6-e6+ 7.Ke2-d2 Qe6xa2+ 8.Kd2-d3 Ke8-e7 9.Kd3-e3 Qa2-c2 10.Ke3-f4 Qc2-e2 11.Kf4-f5 Qe2-e3 12.Kf5-g6 Ke7-e6 13.Kg6-h5 Qe3-g3 14.Kh5-h6 Ke6-f6 15.Kh6-h5 Qg3-h3# 
  -+  (-#15)   Depth: 44/46   00:02:48  384mN, tb=2806124
1...f2-f1Q 2.c7-c8N+ Kd6-d7 3.Ke3xd2 Qf1-a6 4.Nc8-b6+ Qa6xb6 5.e7-e8R Kd7xe8 6.Kd2-e2 Qb6-e6+ 7.Ke2-d2 Qe6xa2+ 8.Kd2-d3 Ke8-e7 9.Kd3-e3 Qa2-c2 10.Ke3-f4 Qc2-e2 11.Kf4-f5 Qe2-e3 12.Kf5-g6 Ke7-e6 13.Kg6-h5 Qe3-g3 14.Kh5-h6 Ke6-f6 15.Kh6-h5 Qg3-h3# 
  -+  (-#15)   Depth: 45/46   00:03:04  420mN, tb=3134809
1...d2-d1Q 2.c7-c8N+ Kd6-c5 3.Ke3xf2 Qd1-d7 4.Kf2-e3 Qd7xc8 5.a2-a4 Qc8-h3+ 6.Ke3-d2 Qh3-d7+ 7.Kd2-e3 Qd7xe7+ 8.Ke3-d3 Kc5-d5 9.a4-a5 Qe7-a3+ 10.Kd3-e2 Kd5-e4 11.Ke2-f2 Qa3-f3+ 12.Kf2-g1 Qf3-g3+ 13.Kg1-h1 Ke4-f3 14.a5-a6 Qg3-g2# 
  -+  (-#14)   Depth: 46/46   00:04:17  593mN, tb=4143001
1...d2-d1Q 2.c7-c8N+ Kd6-c5 3.Ke3xf2 Qd1-d7 4.Kf2-e3 Qd7xc8 5.a2-a4 Qc8-h3+ 6.Ke3-d2 Qh3-d7+ 7.Kd2-e3 Qd7xe7+ 8.Ke3-d3 Kc5-d5 9.a4-a5 Qe7-a3+ 10.Kd3-e2 Kd5-e4 11.Ke2-f2 Qa3-f3+ 12.Kf2-g1 Qf3-g3+

mcostalba · Post by **mcostalba** » Fri Jun 03, 2016 7:07 am

petero2 wrote:
bob wrote:(1) sacrificing your queen takes you into a won 6 piece ending, but one which is harder to win than keeping the queen on. Unfortunately, you can't search nearly deep enough to see the win with the queen kept on the board, so transitioning to the egtb win is the only winning option you can see;
Do you have an example of such a position? It seems to me that if you are enough ahead to still be able to win after a queen sacrifice, you will also often be able to force a more advantageous trade if you keep searching.

The hardest position I could find after arbitrarily trying some positions was this:
[d]8/4bk2/4br2/8/8/8/8/QRRK4 w - - 0 1
Even in this position it only takes texel about 2.5 seconds to find something better than the queen sacrifice:

For Natural TB queen sacrifice is never an option (even at shallowest depths), insetad it finds rather quickly mate in 14

Code: Select all

  +-  (#14)   Depth: 32/84   00:01:16  143mN, tb=1957882
1.Rc1-c7 Be6-g4+ 2.Kd1-d2 Rf6-d6+ 3.Kd2-c3 Rd6-e6 4.Rb1-f1+ Kf7-g6 5.Qa1-b1+ Kg6-h6 6.Rc7xe7 Re6xe7 7.Rf1-f6+ Kh6-g5 8.Qb1-g6+ Kg5-h4 9.Qg6-h6+ Kh4-g3 10.Qh6-f4+ Kg3-h3 11.Rf6-h6+ Bg4-h5 12.Rh6xh5+ Kh3-g2 13.Rh5-g5+ Kg2-h3 14.Qf4-g3#

How much it takes Texel to find mate in 14?

jhellis3 · Post by **jhellis3** » Fri Jun 03, 2016 8:17 am

Very nice Marco, good work with the early draw detection and return too

.

Evert · Post by **Evert** » Fri Jun 03, 2016 8:21 am

syzygy wrote:
Evert wrote:But a Multi-PV like approach seems more reasonable, and fits better with what one could already do for analysis purposes.
Multi-PV is not going to help where the only win that the engine can see is by a conversion into TBs. Basically the engine will need to decide that it does not want the TB win...

That's why I said Multi-PV like. The engine would need to operate in a mode where it will consider the TB result as a type of bound, to answer the question: ok, I found this conversion to a won TB position by tossing my Queen. Is there a faster/alternate win if I don't do that? This being up to the search, it may not find it, so the conversion to the TB win stands.

mcostalba · Post by **mcostalba** » Fri Jun 03, 2016 8:59 am

jhellis3 wrote:Very nice Marco, good work with the early draw detection and return too .

Thanks!

We achieve same result in different ways, you with a smart blending of TB and evaluation scores, I with a new(*) approach to TB usage in the search.

(*) I am not sure it is new, but when the usual "scholars" label it: absurd, ilogic, useless, etc. for me it is comforting and a sign that I am really doing something new.

Scholars just understand legacy and always lack behind new developments (never anticipate or foresee them), sometime even many years later...

Michel · Post by **Michel** » Fri Jun 03, 2016 9:09 am

That's why I said Multi-PV like. The engine would need to operate in a mode where it will consider the TB result as a type of bound, to answer the question: ok, I found this conversion to a won TB position by tossing my Queen. Is there a faster/alternate win if I don't do that? This being up to the search, it may not find it, so the conversion to the TB win stands.

Well this aspect does not need multipv with Joseph Ellis's approach to adjust the TB score with the current evaluation.

syzygy · Post by **syzygy** » Fri Jun 03, 2016 9:41 am

Evert wrote:
syzygy wrote:
Evert wrote:But a Multi-PV like approach seems more reasonable, and fits better with what one could already do for analysis purposes.
Multi-PV is not going to help where the only win that the engine can see is by a conversion into TBs. Basically the engine will need to decide that it does not want the TB win...
That's why I said Multi-PV like. The engine would need to operate in a mode where it will consider the TB result as a type of bound, to answer the question: ok, I found this conversion to a won TB position by tossing my Queen. Is there a faster/alternate win if I don't do that? This being up to the search, it may not find it, so the conversion to the TB win stands.

Yes, and then it still makes the "unnatural" move. To avoid that, the engine will somehow have to decide that it does not want the TB win: "My position is so good that there must be a win that does not involve sacking my queen". That should be possible to implement, but of course it will then fail to win won games in certain circumstances.

Evert · Post by **Evert** » Fri Jun 03, 2016 10:16 am

syzygy wrote: Yes, and then it still makes the "unnatural" move. To avoid that, the engine will somehow have to decide that it does not want the TB win: "My position is so good that there must be a win that does not involve sacking my queen". That should be possible to implement, but of course it will then fail to win won games in certain circumstances.

Of course you play the "unnatural" move if you don't find an alternate win; I don't think anyone argues doing anything different.
If you don't look for an alternative win, you're sure not to find it. That you look for it doesn't mean that you will find it (assuming it does exists).

Again though, I think it's only remotely relevant for analysis. If you just want to win the game and don't care how, none of these cosmetic effects matter.

I think there was a discussion a while back about mate-distance pruning. Does it serve a purpose? Yes, but it doesn't change the outcome of the game. Does it add Elo? No, because it doesn't change the outcome of the game. This is not so different.

Michel · Post by **Michel** » Fri Jun 03, 2016 11:01 am

Yes, and then it still makes the "unnatural" move. To avoid that, the engine will somehow have to decide that it does not want the TB win: "My position is so good that there must be a win that does not involve sacking my queen". That should be possible to implement, but of course it will then fail to win won games in certain circumstances.

As I have explained the engine doesn't have to go that far. It can conceivable just postpone the sacrifice, if the search shows that the sacrifice can still be done later.

The PV will still be ugly (it will contain the sacrifice), but the sacrifice is not made on the board. Of course doing this is pointless if there is not some other measure of progress, otherwise the engine will just play randomly until it runs out of options by the 50move rule and then it will make the sacrifice anyway.

Michel · Post by **Michel** » Fri Jun 03, 2016 3:43 pm

Scholars just understand legacy and always lack behind new developments (never anticipate or foresee them), sometime even many years later...

Remember when you were pushing for dynamic contempt in SF as a magic bullet?

When people said caution was required (extra ordinary claims require extra ordinary proof) you made similar comments.

But in the end dynamic contempt didn't work did it?

Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB

Re: Natural TB