Breakthrough Draughts

[BertTuyt]

Gerard, and here 6x7, seems ok.

Bert

Code: Select all

6x7 P = 90.774.659.360, Capture = 81.674.394.102
6x7 P = 90.774.659.360, CW NCW CL NCL = 45.922.584.894 6.551.349.681 35.751.809.208 2.548.915.577
28424.5 6x7, P = 90.774.659.360, E = 0

[TAILLE]

Gerard, and here 6x7, seems ok.

Bert

Code: Select all

6x7 P = 90.774.659.360, Capture = 81.674.394.102
6x7 P = 90.774.659.360, CW NCW CL NCL = 45.922.584.894 6.551.349.681 35.751.809.208 2.548.915.577
28424.5 6x7, P = 90.774.659.360, E = 0

Good news for me. I just found the cause of my bug on the 12x1 db. A stupid dimensionning error of an index table (76717266 instead of 76717267 !). The last value of the table is used only by the 12xn db when the 12pieces are on the 1-12 squares. As a consequences all 12xn db are wrong.

[BertTuyt]

Gerard, also 7x6 seems good....

Bert

Code: Select all

7x6 P = 90.774.659.360, Capture = 81.537.221.872
7x6 P = 90.774.659.360, CW NCW CL NCL = 61.213.634.774 8.381.897.748 20.323.587.098 855.539.740
24501.3 7x6, P = 90.774.659.360, E = 0

[BertTuyt]

Gerard, and some other 13P results, I did not had the time to compare.

Bert

Code: Select all

8x5 P = 67.013.445.840, Capture = 59.224.793.515
8x5 P = 67.013.445.840, CW NCW CL NCL = 49.547.696.680 7.569.127.412 9.677.096.835 219.524.913
17180.4 8x5, P = 67.013.445.840, E = 0
9x4 P = 36.064.560.004, Capture = 30.692.384.522
9x4 P = 36.064.560.004, CW NCW CL NCL = 27.087.356.350 5.306.714.075 3.605.028.172 65.461.407
8836.7 9x4, P = 36.064.560.004, E = 0
10x3 P = 13.747.443.160, Capture = 10.824.880.467
10x3 P = 13.747.443.160, CW NCW CL NCL = 9.918.390.975 2.903.179.449 906.489.492 19.383.244
3203.9 10x3, P = 13.747.443.160, E = 0
11x2 P = 3.512.903.160, Capture = 2.342.003.640
11x2 P = 3.512.903.160, CW NCW CL NCL = 2.221.419.962 1.166.313.100 120.583.678 4.586.420
779.0 11x2, P = 3.512.903.160, E = 0
12x1 P = 538.899.660, Capture = 235.350.720
12x1 P = 538.899.660, CW NCW CL NCL = 235.350.720 302.994.918 0 554.022
96.5 12x1, P = 538.899.660, E = 0

[TAILLE]

Gerard, and some other 13P results, I did not had the time to compare.

Bert

Code: Select all

8x5 P = 67.013.445.840, Capture = 59.224.793.515
8x5 P = 67.013.445.840, CW NCW CL NCL = 49.547.696.680 7.569.127.412 9.677.096.835 219.524.913
17180.4 8x5, P = 67.013.445.840, E = 0
9x4 P = 36.064.560.004, Capture = 30.692.384.522
9x4 P = 36.064.560.004, CW NCW CL NCL = 27.087.356.350 5.306.714.075 3.605.028.172 65.461.407
8836.7 9x4, P = 36.064.560.004, E = 0
10x3 P = 13.747.443.160, Capture = 10.824.880.467
10x3 P = 13.747.443.160, CW NCW CL NCL = 9.918.390.975 2.903.179.449 906.489.492 19.383.244
3203.9 10x3, P = 13.747.443.160, E = 0
11x2 P = 3.512.903.160, Capture = 2.342.003.640
11x2 P = 3.512.903.160, CW NCW CL NCL = 2.221.419.962 1.166.313.100 120.583.678 4.586.420
779.0 11x2, P = 3.512.903.160, E = 0
12x1 P = 538.899.660, Capture = 235.350.720
12x1 P = 538.899.660, CW NCW CL NCL = 235.350.720 302.994.918 0 554.022
96.5 12x1, P = 538.899.660, E = 0

8x5, 9x4, 10x3 and 11x2 identical results

[TAILLE]

12x1 results are now identical to yours
12x1 NCW1= 33 476 610 NCL1= 245 545 002 NCW2= 302 994 918 NCL2= 554 022
I expect I will able to give you this evening all results for the 14p db.

[TAILLE]

Hi Bert,

First figures for the 14p db
12x2 NCW1= 302 705 258 NCL1= 1 008 172 770 NCW2= 1 484 410 911 NCL2= 3 938 613
11x3 NCW1= 1 227 707 632 NCL1= 2 320 901 210 NCW2= 3 990 227 861 NCL2= 17 199 718
10x4 NCW1= 3 283 778 890 NCL1= 3 909 798 894 NCW2= 7 861 805 321 NCL2= 58 604 100

[BertTuyt]

Gerard, Im again back in Switzerland.

Will first implement a parallel Verify (as my computer here is not so fast).
Anyway my non-compressed figures 2P-13P 76.8 GByte, 2P - 14P 190 GByte...

Hope to provide some results this weekend.

Bert

[TAILLE]

Hi Bert,

Good news.
I did a complete analysis of the start position with the help of 2-14p db + 12x3db + 11x4db + 12x4db.
It took 3h18' and Damy found two winning moves (22-17 and 22-18) and five losing moves.
As you see Damy confirms now that the 22-18 move you found earlier is a winning move.

10x5db is under generation

[TAILLE]

Hi Bert,

Did you developped a perft founction for 8x8 breakthrough game?
I did but, due to my special format of the positions, and doubt we can easliy exchange on that point.

[BertTuyt]

Gerard, no I did not develop a Perft() for BT Draughts.
Maybe we need to do, but as the 13P DB seems to be ok, I don't see much added value.
The results regarding my normal MoveGenerator() are recently posted in the Perft post.

Bert

[jj]

Hi Bert,

Did you developped a perft founction for 8x8 breakthrough game?
I did but, due to my special format of the positions, and doubt we can easliy exchange on that point.

This is what I get for 8x8 breakthrough draughts. The first 7 ply are identical to normal 8x8 draughts.

Code: Select all

perft(1) 7 nodes 1 msec 7 kN/s
perft(2) 49 nodes 1 msec 49 kN/s
perft(3) 302 nodes 1 msec 302 kN/s
perft(4) 1469 nodes 2 msec 734 kN/s
perft(5) 7473 nodes 3 msec 2491 kN/s
perft(6) 37628 nodes 4 msec 9407 kN/s
perft(7) 187302 nodes 16 msec 11706 kN/s
perft(8) 907788 nodes 45 msec 20173 kN/s
perft(9) 4406111 nodes 86 msec 51233 kN/s
perft(10) 21320471 nodes 309 msec 68998 kN/s
perft(11) 103552081 nodes 1492 msec 69404 kN/s
perft(12) 498338018 nodes 7066 msec 70526 kN/s
perft(13) 2387390847 nodes 34578 msec 69043 kN/s
perft(14) 11403507914 nodes 162706 msec 70086 kN/s
perft(15) 54508477664 nodes 794168 msec 68635 kN/s

This is from my 8x8 bitboard perft, with one line added (if #kings > 0 return 1). Up to and including 13 ply confirmed by my general program for WxH boards (see viewtopic.php?f=53&t=4563#p120132).

Jan-Jaap

[Rein Halbersma]

Hi Bert,

Did you developped a perft founction for 8x8 breakthrough game?
I did but, due to my special format of the positions, and doubt we can easliy exchange on that point.

This is what I get for 8x8 breakthrough draughts. The first 7 ply are identical to normal 8x8 draughts.

Code: Select all

perft(13) 2387390847 nodes 34578 msec 69043 kN/s
perft(14) 11403507914 nodes 162706 msec 70086 kN/s
perft(15) 54508477664 nodes 794168 msec 68635 kN/s

This is from my 8x8 bitboard perft, with one line added (if #kings > 0 return 1). Up to and including 13 ply confirmed by my general program for WxH boards (see viewtopic.php?f=53&t=4563#p120132).

Jan-Jaap

Confirmed! (dropping duplicate captures)

Code: Select all

info depth 13 leafs   2387390847 time  16143 147890 knps
info depth 14 leafs  11403507914 time  77397 147338 knps
info depth 15 leafs  54508477664 time 366655 148664 knps

[TAILLE]

Hi Bert,

Did you developped a perft founction for 8x8 breakthrough game?
I did but, due to my special format of the positions, and doubt we can easliy exchange on that point.

This is what I get for 8x8 breakthrough draughts. The first 7 ply are identical to normal 8x8 draughts.

Code: Select all

perft(13) 2387390847 nodes 34578 msec 69043 kN/s
perft(14) 11403507914 nodes 162706 msec 70086 kN/s
perft(15) 54508477664 nodes 794168 msec 68635 kN/s

This is from my 8x8 bitboard perft, with one line added (if #kings > 0 return 1). Up to and including 13 ply confirmed by my general program for WxH boards (see viewtopic.php?f=53&t=4563#p120132).

Jan-Jaap

Confirmed! (dropping duplicate captures)

Code: Select all

info depth 13 leafs   2387390847 time  16143 147890 knps
info depth 14 leafs  11403507914 time  77397 147338 knps
info depth 15 leafs  54508477664 time 366655 148664 knps

As expected my figures are very different due to to my special format
Anyway one sgnificant result:
perft(13) 495 970 174 leaves 20s

[BertTuyt]

Gerard, 2x12 confirmed ok.

Bert

Code: Select all

Verify:2x12 Start, P = 4.516.906.290
839.1 2x12, P = 4.516.906.290, NCW = 302.705.258, E = 0