Perft(14) estimates thread

Discussion of chess software programming and technical issues.

Moderators: hgm, Rebel, chrisw

ibid
Posts: 89
Joined: Mon Jun 13, 2011 12:09 pm

Re: Smallest perft(10) sub-result (?)

Post by ibid »

petero2 wrote:
sje wrote:Smallest perft(10) sub-result (?)
[d]rnbqkbnr/pp1pp1pp/2p2p2/8/8/2P2P2/PP1PP1PP/RNBQKBNR w KQkq - 0 3[/d]
That is not the smallest. This is smaller:
[d]rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 4 3[/d]
I suppose in principle something like this wins:
[d]rnb1kbnr/pppp1ppp/4p3/8/5PPq/8/PPPPP2P/RNBQKBNR w KQkq -[/d]
A somewhat more reasonable position:
[d]rnb1kbnr/pppp1ppp/4p3/8/5P1q/8/PPPPP1PP/RNBQKBNR w KQkq -[/d]
has a perft(10) of 17,401,317,066,205. I wonder at what point the extra moves from the active black queen become more important than the missing ply due to the check?
petero2
Posts: 683
Joined: Mon Apr 19, 2010 7:07 pm
Location: Sweden
Full name: Peter Osterlund
Contact:

Re: Perft(14) estimates thread

Post by petero2 »

My perft(14) calculation is now finished and the result is:

61,885,021,521,585,529,237

It seems reasonable given earlier estimates, but obviously needs independent verification.
User avatar
Ajedrecista
Posts: 1966
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.
Contact:

Perft(14) completed!

Post by Ajedrecista »

Hello Peter:
petero2 wrote:My perft(14) calculation is now finished and the result is:

61,885,021,521,585,529,237

It seems reasonable given earlier estimates, but obviously needs independent verification.
Congratulations for this huge task!

There are some estimates, some of them quite old:

Code: Select all

61,803,489,628,662,504,195 by Joshua Haglund.

61,865,545,676,422,373,910 by Matthew R. Brades. 

6.187e+19 by François Labelle.

6.18847822079403e+19 by Peter Österlund. 

61,886,459,822,115,294,738 by myself. 

6.188925e+19 by H.G.Muller.

6.18986e+19 by Reinhard Scharnagl.
If I am not wrong, these are the relative errors err_i = [estimate_i/Perft(14) - 1] rounding up to 0.0001%:

Code: Select all

    Joshua Haglund's estimate ~ Perft(14) - 0.1317 % ; |err| ~ 1/(759.03)

  Matthew R. Brades' estimate ~ Perft(14) - 0.0315 % ; |err| ~ 1/(3177.53)

  François Labelle's estimate ~ Perft(14) - 0.0243 % ; |err| ~ 1/(4119.76)

   Peter Österlund's estimate ~ Perft(14) - 0.0004 % ; |err| ~ 1/(258593.79)

                  My estimate ~ Perft(14) + 0.0023 % ; |err| ~ 1/(43026.49)

        H.G.Muller's estimate ~ Perft(14) + 0.0068 % ; |err| ~ 1/(14635.29)

Reinhard Scharnagl's estimate ~ Perft(14) + 0.0219 % ; |err| ~ 1/(4557.58)
I hope no typos.
Ajedrecista wrote:I expect to be wrong in less than 0.01% this time. It is too soon but I think that Peter's estimate is the best one, followed by my own estimate.
I was far below 0.01 % of relative error! But I rightly wrote that Peter's estimate would be the closest one. Indeed it is superb... once again, just like with his Perft(13) estimate. Well done!

Waiting for independent verification... it will take a while.

Regards from Spain.

Ajedrecista.
User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 7:43 pm

Re: Perft(14) estimates thread

Post by sje »

petero2 wrote:My perft(14) calculation is now finished and the result is:

61,885,021,521,585,529,237
Congratulations. Now we need the 400 draft 12 results for a check. These should be done on a different machine by a different party with a different algorithm.
jhaglund
Posts: 173
Joined: Sun May 11, 2008 7:43 am

Re: Perft(14) completed!

Post by jhaglund »

Ajedrecista wrote:Hello Peter:
petero2 wrote:My perft(14) calculation is now finished and the result is:

61,885,021,521,585,529,237

It seems reasonable given earlier estimates, but obviously needs independent verification.
Congratulations for this huge task!

There are some estimates, some of them quite old:

Code: Select all

61,803,489,628,662,504,195 by Joshua Haglund.

61,865,545,676,422,373,910 by Matthew R. Brades. 

6.187e+19 by François Labelle.

6.18847822079403e+19 by Peter Österlund. 

61,886,459,822,115,294,738 by myself. 

6.188925e+19 by H.G.Muller.

6.18986e+19 by Reinhard Scharnagl.
If I am not wrong, these are the relative errors err_i = [estimate_i/Perft(14) - 1] rounding up to 0.0001%:

Code: Select all

    Joshua Haglund's estimate ~ Perft(14) - 0.1317 % ; |err| ~ 1/(759.03)

  Matthew R. Brades' estimate ~ Perft(14) - 0.0315 % ; |err| ~ 1/(3177.53)

  François Labelle's estimate ~ Perft(14) - 0.0243 % ; |err| ~ 1/(4119.76)

   Peter Österlund's estimate ~ Perft(14) - 0.0004 % ; |err| ~ 1/(258593.79)

                  My estimate ~ Perft(14) + 0.0023 % ; |err| ~ 1/(43026.49)

        H.G.Muller's estimate ~ Perft(14) + 0.0068 % ; |err| ~ 1/(14635.29)

Reinhard Scharnagl's estimate ~ Perft(14) + 0.0219 % ; |err| ~ 1/(4557.58)
I hope no typos.
Ajedrecista wrote:I expect to be wrong in less than 0.01% this time. It is too soon but I think that Peter's estimate is the best one, followed by my own estimate.
I was far below 0.01 % of relative error! But I rightly wrote that Peter's estimate would be the closest one. Indeed it is superb... once again, just like with his Perft(13) estimate. Well done!

Waiting for independent verification... it will take a while.

Regards from Spain.

Ajedrecista.

You guys are too quick, I didn't get time to update my prediction from, Aug 20, 2011. :lol:

Good work! Are we looking into Peftf(15) ?
User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 7:43 pm

Re: Perft(14) estimates thread

Post by sje »

sje wrote:
petero2 wrote:My perft(14) calculation is now finished and the result is:

61,885,021,521,585,529,237
Congratulations. Now we need the 400 draft 12 results for a check. These should be done on a different machine by a different party with a different algorithm.
I'll contribute:
[d]rnbqkbnr/ppppp1pp/5p2/8/8/5P2/PPPPP1PP/RNBQKBNR w KQkq - 0 2[/d]

Code: Select all

Depth: 12   Count: 32,805,467,443,838,610   Elapsed: 526411  (6.23192e+10 Hz / 1.60464e-11 s)
Post Reply