I have computed perft(17)

[Ajedrecista]

Hello:

I ask/answer various persons in the same post:

------------------------

I have computed perft(17) on a few million core-hours access I had to a supercomputer at Aragonne labs. I used close to 700000 cores for computation so the computational power is really big. Those guys will not allow you to waste their resources without purpose and benefit to the society, so I had to write a proposal and take training to get that done. Anyway I am not sure if everything went well as intended, so I can not be sure if the result will hold when I do the verification later. Without further adieu, here is the result I got for perft(17) that is not verified yet...drum rolls please...
2172314159265358979323846

Why do not compute first Perft(14), Perft(15) and Perft(16)? It looks very strange for me. I also did not know that you have written a multi-core perft utility for Nebiyu. Could you provide the CPU time and the real-life time of the alleged calculation, please? Thanks in advance.

------------------------

What program did you use? And by the way, your calculation suggests perft(17) had a branching factor of almost 27, so it seems somewhat correct.

Please take a look here:

Statistics on chess games

Code: Select all

Perft estimates
 n	 perft(n)
13	1.9810e+18
14	6.187e+19
15	2.015e+21
16	6.507e+22
17	2.175e+24
18	7.214e+25
19	2.462e+27
20	8.350e+28

The estimate of Perft(17) does not differ a lot from the alleged result. How one can be safe that knowing that estimate (or others that are in line), someone pounds the keyboard maintaining the first three or four digits, then going until the 25th? Sorry Daniel, but this is a question that many people could be wondering now... I know your efforts with MonteCarlo perft simulations in the past.

There are other estimates here in TalkChess:

Re: Perft(20) summary of estimates

Code: Select all

perft(17)= ca.2.172185e+24 (74.900 sec)  30.983098 (Müller, 2011/07/10; the last number is the branching factor).

Perft(17) ~ 2.1729e+24 (Muñoz, 2011/08/04; myself with my hocus-pocus method of logarithms, polynomial adjusts, Lagrange polinomials, etc., explained in some old posts by myself).

perft(17) estimate (1,000,000 random walks in about 78 seconds): 2.1782e+24 (Schüle using Österlund's method, 2011/08/09).

------------------------

Well, if it's not already getting off topic, I've an good approximation for pi:

Code: Select all

      426880 * Sqrt(10005)
pi ~= --------------------
            13591409

I discovered by chance a fairly good approximation to pi some years ago (probably when I was 14 or 15) with only two 3's and two 5's:

Code: Select all

3*[3 + sqrt(5)]/5 ~ 3.14164079

|3*[3 + sqrt(5)]/(5*pi) - 1| ~ 1/(65269.17)

|3*[3 + sqrt(5)]/(5*pi) - 1|*100 ~ 0.001532% of error.

I say fairly good given the simplicity of the used numbers.

------------------------

http://en.wikipedia.org/wiki/April_Fools%27_Day

The end of the joke. I was expecting a joke today in the morning but not now... it seems that I am more customed to a similar day in Spain: 28th of December, Día de los Santos Inocentes or simply Día de los Inocentes. The jokes on that day are known as inocentadas:

Día de los Santos Inocentes

You can click on the English article, of course!

------------------------

Regards from Spain.

Ajedrecista.

[Daniel Shawul]

Louii is such a buzz kill

[Ajedrecista]

Hello Daniel:

Louis is such a buzz kill

Not only Louis but also Michael, who was faster!

Cool ! I would repeat the test exactly in one year. Guess that is the best chance for verification Laughing Wink

It was too late for edit my post when I realized on that. Michael was more subtle than Louis.

It was nice to refresh old estimates from year 2011.

Regards from Spain.

Ajedrecista.

[sje]

I have computed perft(17) on a few million core-hours access I had to a supercomputer at Aragonne labs. I used close to 700000 cores for computation so the computational power is really big. Those guys will not allow you to waste their resources without purpose and benefit to the society, so I had to write a proposal and take training to get that done. Anyway I am not sure if everything went well as intended, so I can not be sure if the result will hold when I do the verification later. Without further adieu, here is the result I got for perft(17) that is not verified yet...drum rolls please...
2172314159265358979323846

I've done even better and have computed perft(π) = 10,000 exactly.

[kinderchocolate]

I've solved chess on my mobile phone.

[zullil]

Louii is such a buzz kill

Sorry. I guessed that Syed was totally unfamiliar with the April 1 tradition, so I felt sorry for him and decided to help.

[Daniel Shawul]

Louii is such a buzz kill

Sorry. I guessed that Syed was totally unfamiliar with the April 1 tradition, so I felt sorry for him and decided to help.

No need to apologize. I was hoping someone would stop me anyway. I am starting to realize it is a cruel joke and I should stop atleast until ... next year.
No no scratch that, I really shouldn't be making this pranks ever again.

[Daniel Shawul]

I have computed perft(17) on a few million core-hours access I had to a supercomputer at Aragonne labs. I used close to 700000 cores for computation so the computational power is really big. Those guys will not allow you to waste their resources without purpose and benefit to the society, so I had to write a proposal and take training to get that done. Anyway I am not sure if everything went well as intended, so I can not be sure if the result will hold when I do the verification later. Without further adieu, here is the result I got for perft(17) that is not verified yet...drum rolls please...
2172314159265358979323846

I've done even better and have computed perft(π) = 10,000 exactly.

Now that you have finally posted, it is time for me to anounce, what is obvious by now,
that the figure I gave contains the first 3 digist os a monte-carlo perft estimate of perft(17) + random digit + Digits of Pi
217...2...314159265358979323846

[vittyvirus]

I have computed perft(17) on a few million core-hours access I had to a supercomputer at Aragonne labs. I used close to 700000 cores for computation so the computational power is really big. Those guys will not allow you to waste their resources without purpose and benefit to the society, so I had to write a proposal and take training to get that done. Anyway I am not sure if everything went well as intended, so I can not be sure if the result will hold when I do the verification later. Without further adieu, here is the result I got for perft(17) that is not verified yet...drum rolls please...
2172314159265358979323846

I've done even better and have computed perft(π) = 10,000 exactly.

Now that you have finally posted, it is time for me to anounce, what is obvious by now,
that the figure I gave contains the first 3 digist os a monte-carlo perft estimate of perft(17) + random digit + Digits of Pi
217...2...314159265358979323846

Poor me. Memorizing first 50 digits of pi didn't help me.

[Daniel Shawul]

Sorry Syed, what a coincidence that you happen to be enthusiastic about pi. I just couldn't resist myself from trolling you more