Perft(13), final fifth

[sje]

After 48 weeks:

The run has produced 714,517 draft 8 records, about 87% of the total needed. The average draft 8 record generation time is about 40.6 seconds.

There are now 266 draft 11 records, 66.5% of the 400 needed. The average draft 11 record generation time is about 30.3 hours.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft11

There are now 10 draft 12 records, 50% of the 20 needed. The average draft 12 record generation time is about 33.6 days.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft12

Recent output:

Code: Select all

r1bqkbnr/pppp1ppp/n7/4p3/3P1B2/6P1/PPP1PP1P/RN1QKBNR b KQkq - 2 3 8 1053277691219
rnb1kbnr/pppqpppp/B7/3p4/3P4/4P3/PPP2PPP/RNBQK1NR b KQkq - 2 3 8 1065157686526
rn1qkbnr/pppbpppp/8/3p4/4P1P1/3B4/PPPP1P1P/RNBQK1NR b KQkq - 2 3 8 446597995767
rnb1kbnr/pppp1ppp/8/4p1q1/2P3P1/4P3/PP1P1P1P/RNBQKBNR b KQkq - 0 3 8 1060796750312
rn1qkbnr/pppbpppp/8/3p4/2B1P1P1/8/PPPP1P1P/RNBQK1NR b KQkq - 2 3 8 641611447539
rnb1kbnr/pppqpppp/8/3p4/3P4/4P3/PPP1NPPP/RNBQKB1R b KQkq - 2 3 8 560561706297
rnbqkbnr/ppppp1p1/7B/5p2/8/2PP4/PP2PPPP/RN1QKBNR b KQkq - 0 3 8 292983790658
rn1qkbnr/pppbpppp/8/1B1p4/4P1P1/8/PPPP1P1P/RNBQK1NR b KQkq - 2 3 8 367393983084
rnbqkbnr/ppppp1p1/7p/5p2/8/2PP4/PP1QPPPP/RNB1KBNR b KQkq - 1 3 8 143548114915
rn1qkbnr/pppbpppp/B7/3p4/4P1P1/8/PPPP1P1P/RNBQK1NR b KQkq - 2 3 8 509767355447

[sje]

For the perft(11) of the position after 1 e4 e5, here are eight of the fourteen draft 10 pawn moves:

Code: Select all

a3 630,283,717,546,506
b3 753,067,349,940,642
b4 698,965,944,252,507
a4 827,889,685,802,105
c3 722,535,988,666,500
c4 598,893,210,923,167
f3 329,925,944,246,476
d3 872,864,801,477,844

Six more (pawn) moves to go; maybe another day or two. Note that al of the calculations for this draft 11 number are done with a 56 bit hash and not a 120 bit hash as used in the perft(13) run. It will be interesting to see if there are any differences.

[sje]

For the perft(11) of the position after 1 e4 e5, here are last six of the fourteen draft 10 pawn moves:

Code: Select all

g3 710,121,597,439,690
g4 506,540,698,647,738
f4 768,743,543,843,060
h3 589,069,745,938,800
d4 2,006,488,639,287,825
h4 825,439,188,415,311

The perft(11) total of the position after 1 e4 e5 is: 22,653,878,987,990,237. Again, this was calculated with only 56 bits of hash. The run took just over eleven days of wall time using a 2.66 GHz four core machine.

Within a couple of months, the main perft(13) run should complete and its 120 bit hash result for this position can then be compared.

[sje]

The main run has completed perft(11) for the position after 1 d4 d5. The value is 17,022,349,302,490,272.

[sje]

After 49 weeks:

The run has produced 722,379 draft 8 records, about 88% of the total needed. The average draft 8 record generation time is about 41.0 seconds.

There are now 270 draft 11 records, 67.5% of the 400 needed. The average draft 11 record generation time is about 30.5 hours.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft11

There are now 10 draft 12 records, 50% of the 20 needed. The average draft 12 record generation time is about 34.3 days.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft12

Recent output:

Code: Select all

rnb1kbnr/pppp1ppp/7B/4p1q1/8/3P4/PPP1PPPP/RN1QKBNR w KQkq - 2 3 9 27110316434929
rnbqkbnr/ppp1pppp/8/8/2p1P2P/8/PPPP1PP1/RNBQK1NR b KQkq - 0 3 8 729447012712
rnb1kbnr/pppp1ppp/B7/4p1q1/4P3/6P1/PPPP1P1P/RNBQK1NR b KQkq - 2 3 8 1169195023409
r1bqkbnr/ppppp1pp/2n2p2/8/Q1P1P3/8/PP1P1PPP/RNB1KBNR b KQkq - 0 3 8 316918584889
rnbqkbnr/ppp1pppp/8/4P3/2p5/8/PPPP1PPP/RNBQK1NR b KQkq - 0 3 8 537877908841
rnbqkbnr/ppp1pppp/8/8/2p1P3/8/PPPP1PPP/RNBQK1NR w KQkq - 0 3 9 19224072634378
r1bqkbnr/pppppp1p/2n3p1/8/4P3/2P4P/PP1P1PP1/RNBQKBNR b KQkq - 0 3 8 369197320190
rn1qkbnr/ppp1pppp/8/3p4/7P/3BP2b/PPPP1PP1/RNBQK1NR b KQkq - 2 3 8 1141006417499
rnbqk1nr/pppp1ppp/4p3/8/3P2P1/b6P/PPP1PP2/RNBQKBNR b KQkq - 0 3 8 611158285120
r1bqkbnr/ppppp1pp/2n2p2/8/Q1P5/6P1/PP1PPP1P/RNB1KBNR b KQkq - 0 3 8 260572058313

[sje]

After 50 weeks:

The run has produced 731,998 draft 8 records, about 89% of the total needed. The average draft 8 record generation time is about 41.3 seconds.

There are now 278 draft 11 records, 69.5% of the 400 needed. The average draft 11 record generation time is about 30.2 hours.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft11

There are now 10 draft 12 records, 50% of the 20 needed. The average draft 12 record generation time is about 35.0 days.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft12

Recent output:

Code: Select all

rnbqkbnr/pppp1pp1/4p3/7p/8/4PQ2/PPPP1PPP/RNBK1BNR b kq - 1 3 8 1763566672958
rnbqkb1r/ppppp1pp/7n/5p2/7P/4P1P1/PPPP1P2/RNBQKBNR b KQkq - 0 3 8 285074711626
rnbqkb1r/ppppp1pp/7n/5p2/4P3/6P1/PPPP1P1P/RNBQKBNR b KQkq - 0 3 8 297128810905
rnbqkb1r/ppppp1pp/7n/5p2/6P1/4P3/PPPP1P1P/RNBQKBNR b KQkq - 0 3 8 244885920837
rnbqkb1r/ppppp1pp/7n/5p2/8/4P1P1/PPPP1P1P/RNBQKBNR w KQkq - 1 3 9 7728574756216
rnbqkbnr/ppppp1pp/8/5p2/8/4P1P1/PPPP1P1P/RNBQKBNR b KQkq - 0 2 10 225373691394344
r1bqkbnr/pppppp1p/n5p1/8/2P5/1QN5/PP1PPPPP/R1B1KBNR b KQkq - 3 3 8 359295477956
rnbqkbnr/ppppppp1/7B/7p/3P4/2P5/PP2PPPP/RN1QKBNR b KQkq - 1 3 8 358322542328
rnbqkbnr/ppp1pp1p/8/3p2p1/4P3/8/PPPPNPPP/RNBQKBR1 b Qkq - 1 3 8 428685878777
rnbqkbnr/pppp1p1p/6p1/4p3/6P1/8/PPPPPPBP/RNBQ1KNR b kq - 1 3 8 541542238199

[sje]

For the position after 1 e4 d5, the perft(11) is 28,808,031,383,517,106 and this may be the largest of all the 400 perft(11) values for the positions two ply from the root.

[Ajedrecista]

Hello Steven:

You are doing a great work, but this is nothing new.

For the position after 1 e4 d5, the perft(11) is 28,808,031,383,517,106 and this may be the largest of all the 400 perft(11) values for the positions two ply from the root.

I disagree with you about the largest draft 11 record of Perft(13): taking a quick look at Perft(11), I realized that there are three larger draft 9 records of Perft(11) than e4, d5: they are e3, e6 (the largest); e4, e6 (the second largest) and e3, e5 (the third largest):

Code: Select all

m1        m2      count
e2e3      e7e6    26960191086033
e2e4      e7e6    26598009679041
e2e3      e7e5    26445262588567
e2e4      d7d5    26153690791449

I think that this pattern will not change a lot in draft 11 records of Perft(13). I have done some short MonteCarlo perft estimates with the priceless help of GNU 5.07.173b w32 and its command line 'perftmc':

Code: Select all

1.- e3, e6; MonteCarlo perft(11):

FEN=rnbqkbnr/pppp1ppp/4p3/8/8/4P3/PPPP1PPP/RNBQKBNR w KQkq - 0 2
white  KQkq
r n b q k b n r
p p p p . p p p
. . . . p . . .
. . . . . . . .
. . . . . . . .
. . . . P . . .
P P P P . P P P
R N B Q K B N R

perftmc 11
m=2.954491e+016 sd=1.188594e+013 ci(99%)=[2.951429e+016,2.957553e+016] n=11597340 sdn=4.047740e+016 t=114.59s
m=2.955449e+016 sd=1.178964e+013 ci(99%)=[2.952412e+016,2.958486e+016] n=17396048 sdn=4.917289e+016 t=171.73s
m=2.955676e+016 sd=8.639515e+012 ci(99%)=[2.953451e+016,2.957902e+016] n=23194782 sdn=4.160874e+016 t=228.52s
m=2.955276e+016 sd=7.796945e+012 ci(99%)=[2.953268e+016,2.957285e+016] n=28993495 sdn=4.198312e+016 t=285.47s
m=2.955355e+016 sd=6.414750e+012 ci(99%)=[2.953702e+016,2.957007e+016] n=34792189 sdn=3.783734e+016 t=342.44s
m=2.955846e+016 sd=7.313945e+012 ci(99%)=[2.953962e+016,2.957730e+016] n=40590861 sdn=4.659784e+016 t=399.42s
m=2.955511e+016 sd=7.164675e+012 ci(99%)=[2.953665e+016,2.957357e+016] n=46389522 sdn=4.879850e+016 t=456.53s
m=2.955484e+016 sd=6.324287e+012 ci(99%)=[2.953855e+016,2.957113e+016] n=52188174 sdn=4.568753e+016 t=513.58s
m=2.955033e+016 sd=7.237408e+012 ci(99%)=[2.953168e+016,2.956897e+016] n=57986866 sdn=5.511221e+016 t=570.52s
m=2.955335e+016 sd=7.208427e+012 ci(99%)=[2.953478e+016,2.957191e+016] n=63785530 sdn=5.757071e+016 t=627.45s
m=2.955245e+016 sd=6.641583e+012 ci(99%)=[2.953534e+016,2.956955e+016] n=69584267 sdn=5.540221e+016 t=684.59s
m=2.955327e+016 sd=6.165020e+012 ci(99%)=[2.953739e+016,2.956915e+016] n=75382991 sdn=5.352679e+016 t=741.58s
m=2.955426e+016 sd=5.791863e+012 ci(99%)=[2.953934e+016,2.956918e+016] n=81181658 sdn=5.218518e+016 t=798.30s
m=2.955469e+016 sd=5.409206e+012 ci(99%)=[2.954075e+016,2.956862e+016] n=86980372 sdn=5.044802e+016 t=855.27s
m=2.955365e+016 sd=5.165632e+012 ci(99%)=[2.954034e+016,2.956695e+016] n=92778991 sdn=4.975632e+016 t=912.36s

Interrupted!

Code: Select all

1.- e4, e6; MonteCarlo perft(11):

FEN=rnbqkbnr/pppp1ppp/4p3/8/4P3/8/PPPP1PPP/RNBQKBNR w KQkq - 0 2
white  KQkq
r n b q k b n r
p p p p . p p p
. . . . p . . .
. . . . . . . .
. . . . P . . .
. . . . . . . .
P P P P . P P P
R N B Q K B N R

perftmc 11
m=2.891130e+016 sd=2.942705e+012 ci(99%)=[2.890372e+016,2.891888e+016] n=11569564 sdn=1.000934e+016 t=114.31s
m=2.890620e+016 sd=5.375176e+012 ci(99%)=[2.889236e+016,2.892005e+016] n=17354317 sdn=2.239218e+016 t=171.67s
m=2.891164e+016 sd=6.637332e+012 ci(99%)=[2.889455e+016,2.892874e+016] n=23139158 sdn=3.192768e+016 t=229.34s
m=2.891386e+016 sd=5.598457e+012 ci(99%)=[2.889944e+016,2.892828e+016] n=28924005 sdn=3.010909e+016 t=286.52s
m=2.891408e+016 sd=4.576389e+012 ci(99%)=[2.890229e+016,2.892587e+016] n=34708806 sdn=2.696142e+016 t=343.52s
m=2.891220e+016 sd=4.299207e+012 ci(99%)=[2.890113e+016,2.892328e+016] n=40493566 sdn=2.735781e+016 t=400.47s
m=2.891546e+016 sd=4.948197e+012 ci(99%)=[2.890271e+016,2.892821e+016] n=46278467 sdn=3.366173e+016 t=457.58s
m=2.891802e+016 sd=5.058324e+012 ci(99%)=[2.890499e+016,2.893105e+016] n=52063216 sdn=3.649826e+016 t=514.69s
m=2.892166e+016 sd=5.808634e+012 ci(99%)=[2.890670e+016,2.893663e+016] n=57848002 sdn=4.417923e+016 t=571.80s
m=2.891806e+016 sd=6.373074e+012 ci(99%)=[2.890164e+016,2.893447e+016] n=63632871 sdn=5.083815e+016 t=628.91s
m=2.891825e+016 sd=5.821186e+012 ci(99%)=[2.890326e+016,2.893325e+016] n=69417584 sdn=4.850050e+016 t=686.00s
m=2.891921e+016 sd=5.439840e+012 ci(99%)=[2.890520e+016,2.893323e+016] n=75202392 sdn=4.717392e+016 t=743.19s
m=2.891757e+016 sd=5.295969e+012 ci(99%)=[2.890393e+016,2.893122e+016] n=80987188 sdn=4.765995e+016 t=800.39s
m=2.891761e+016 sd=4.930411e+012 ci(99%)=[2.890491e+016,2.893031e+016] n=86772029 sdn=4.592752e+016 t=857.55s
m=2.891761e+016 sd=4.611978e+012 ci(99%)=[2.890573e+016,2.892949e+016] n=92556731 sdn=4.437019e+016 t=914.73s

Interrupted!

Code: Select all

1.- e3, e5; MonteCarlo perft(11):

FEN=rnbqkbnr/pppp1ppp/8/4p3/8/4P3/PPPP1PPP/RNBQKBNR w KQkq - 0 2
white  KQkq
r n b q k b n r
p p p p . p p p
. . . . . . . .
. . . . p . . .
. . . . . . . .
. . . . P . . .
P P P P . P P P
R N B Q K B N R

perftmc 11
m=2.869135e+016 sd=1.270672e+013 ci(99%)=[2.865862e+016,2.872409e+016] n=11583924 sdn=4.324754e+016 t=113.66s
m=2.869776e+016 sd=9.741558e+012 ci(99%)=[2.867267e+016,2.872286e+016] n=17375908 sdn=4.060712e+016 t=170.67s
m=2.869993e+016 sd=7.220379e+012 ci(99%)=[2.868133e+016,2.871853e+016] n=23167747 sdn=3.475377e+016 t=227.62s
m=2.869922e+016 sd=5.637370e+012 ci(99%)=[2.868470e+016,2.871374e+016] n=28959685 sdn=3.033706e+016 t=284.59s
m=2.869572e+016 sd=5.785927e+012 ci(99%)=[2.868081e+016,2.871062e+016] n=34751520 sdn=3.410828e+016 t=341.58s
m=2.869092e+016 sd=6.848780e+012 ci(99%)=[2.867328e+016,2.870856e+016] n=40543496 sdn=4.360877e+016 t=398.58s
m=2.868961e+016 sd=6.073380e+012 ci(99%)=[2.867397e+016,2.870526e+016] n=46335300 sdn=4.134152e+016 t=455.55s
m=2.869461e+016 sd=7.328066e+012 ci(99%)=[2.867574e+016,2.871349e+016] n=52127173 sdn=5.290802e+016 t=512.52s
m=2.869353e+016 sd=6.642860e+012 ci(99%)=[2.867642e+016,2.871065e+016] n=57919108 sdn=5.055523e+016 t=569.47s
m=2.869353e+016 sd=6.008694e+012 ci(99%)=[2.867805e+016,2.870901e+016] n=63710999 sdn=4.796090e+016 t=626.45s
m=2.869559e+016 sd=5.860439e+012 ci(99%)=[2.868050e+016,2.871069e+016] n=69502968 sdn=4.885756e+016 t=683.27s
m=2.869544e+016 sd=5.393042e+012 ci(99%)=[2.868155e+016,2.870933e+016] n=75294847 sdn=4.679683e+016 t=740.09s
m=2.869647e+016 sd=5.097843e+012 ci(99%)=[2.868334e+016,2.870960e+016] n=81086809 sdn=4.590517e+016 t=797.14s
m=2.869408e+016 sd=5.314627e+012 ci(99%)=[2.868039e+016,2.870777e+016] n=86878806 sdn=4.953700e+016 t=854.14s
m=2.869331e+016 sd=5.029406e+012 ci(99%)=[2.868036e+016,2.870627e+016] n=92670749 sdn=4.841590e+016 t=911.19s

Interrupted!

Checking the accuracy of these MonteCarlo perft estimates with 1.- e4, d5:

Code: Select all

1.- e4, d5; MonteCarlo perft(11):

FEN=rnbqkbnr/ppp1pppp/8/3p4/4P3/8/PPPP1PPP/RNBQKBNR w KQkq - 0 2
white  KQkq
r n b q k b n r
p p p . p p p p
. . . . . . . .
. . . p . . . .
. . . . P . . .
. . . . . . . .
P P P P . P P P
R N B Q K B N R

perftmc 11
m=2.881235e+016 sd=4.003087e+012 ci(99%)=[2.880203e+016,2.882266e+016] n=11093572 sdn=1.333309e+016 t=110.81s
m=2.880609e+016 sd=6.672581e+012 ci(99%)=[2.878890e+016,2.882328e+016] n=16640394 sdn=2.721922e+016 t=166.17s
m=2.880015e+016 sd=7.584608e+012 ci(99%)=[2.878061e+016,2.881969e+016] n=22187218 sdn=3.572602e+016 t=221.53s
m=2.880204e+016 sd=6.170638e+012 ci(99%)=[2.878614e+016,2.881793e+016] n=27734086 sdn=3.249653e+016 t=276.89s
m=2.879955e+016 sd=5.618191e+012 ci(99%)=[2.878508e+016,2.881402e+016] n=33280864 sdn=3.241110e+016 t=332.08s
m=2.880311e+016 sd=5.937526e+012 ci(99%)=[2.878782e+016,2.881841e+016] n=38827720 sdn=3.699785e+016 t=387.44s
m=2.880080e+016 sd=5.637752e+012 ci(99%)=[2.878628e+016,2.881533e+016] n=44374516 sdn=3.755543e+016 t=442.75s
m=2.879931e+016 sd=5.192187e+012 ci(99%)=[2.878593e+016,2.881268e+016] n=49921331 sdn=3.668541e+016 t=498.09s
m=2.879734e+016 sd=5.044991e+012 ci(99%)=[2.878434e+016,2.881033e+016] n=55468098 sdn=3.757353e+016 t=553.47s
m=2.879598e+016 sd=4.759934e+012 ci(99%)=[2.878372e+016,2.880824e+016] n=61014939 sdn=3.718082e+016 t=608.83s
m=2.879876e+016 sd=5.156774e+012 ci(99%)=[2.878548e+016,2.881204e+016] n=66561703 sdn=4.207172e+016 t=664.17s
m=2.880056e+016 sd=5.074915e+012 ci(99%)=[2.878749e+016,2.881364e+016] n=72108529 sdn=4.309453e+016 t=719.50s
m=2.879992e+016 sd=4.741695e+012 ci(99%)=[2.878771e+016,2.881214e+016] n=77655447 sdn=4.178492e+016 t=774.55s
m=2.879931e+016 sd=4.457373e+012 ci(99%)=[2.878782e+016,2.881079e+016] n=83202277 sdn=4.065805e+016 t=829.84s
m=2.880040e+016 sd=4.310762e+012 ci(99%)=[2.878930e+016,2.881150e+016] n=88749059 sdn=4.061027e+016 t=885.12s

Interrupted!

The last output (2.88004e+16) is around 0.0265% smaller than the true perft value, so these approximations are quite valid. It looks like 1.- e3, e6 and 1.- e4, e6 will have greater perft(11) values than the Scandinavian defence.

Any thoughts are welcome. Please keep up the good work!

Regards from Spain.

Ajedrecista.

[sje]

The program has a worker thread doing 1 e3 e6 and 1 e4 e6 right now.

The worker thread handing the 1 c3 tree is almost done and should finish in a few days. When it completes, the worker thread for 1 h3 will start, leaving only the 1 h4 thread of the 20 ply 1 positions not yet started.

[sje]

After 51 weeks:

The run has produced 740,531 draft 8 records, about 90% of the total needed. The average draft 8 record generation time is about 41.7 seconds.

There are now 284 draft 11 records, 71% of the 400 needed. The average draft 11 record generation time is about 30.2 hours.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft11

There are now 10 draft 12 records, 50% of the 20 needed. The average draft 12 record generation time is about 35.7 days.
http://dl.dropbox.com/u/31633927/Perft/Perft13/draft12

Recent output:

Code: Select all

r1bqkb1r/pppppppp/n6n/8/Q7/2P5/PP1PPPPP/RNB1KBNR w KQkq - 3 3 9 8712193868341
r1bqkbnr/pppppppp/n7/8/Q7/2P5/PP1PPPPP/RNB1KBNR b KQkq - 2 2 10 208109869919884
rnbqkbnr/pppppp1p/8/8/2P3p1/3P4/PP1QPPPP/RNB1KBNR b KQkq - 1 3 8 269443823694
rnbqkb1r/pppppp1p/5n2/6p1/8/4P1PP/PPPP1P2/RNBQKBNR b KQkq - 0 3 8 327773571652
rnbqkbnr/pppppp1p/8/8/2P3p1/3P4/PP1KPPPP/RNBQ1BNR b kq - 1 3 8 149137179187
r1bqkbnr/pppppppp/8/8/1n6/2PP4/PP1NPPPP/R1BQKBNR b KQkq - 2 3 8 222251477920
rnbqkb1r/pppppp1p/5n2/6p1/7P/4P1P1/PPPP1P2/RNBQKBNR b KQkq - 0 3 8 537912846171
rnb1kbnr/pppp1ppp/4p3/6q1/4P1Q1/8/PPPP1PPP/RNB1KBNR b KQkq - 3 3 8 3459245153633
rnbqk1nr/pppp1ppp/7B/2b1p3/3P2P1/8/PPP1PP1P/RN1QKBNR b KQkq - 2 3 8 1059642429410
rnbqkbnr/pppppp1p/8/8/2P3p1/3PP3/PP3PPP/RNBQKBNR b KQkq - 0 3 8 253913513435