Perft(14) Weekly Status Reports for 2016

Discussion of chess software programming and technical issues.

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Perft(14) Weekly Status 2016-05-29

Post by sje » Sun May 29, 2016 6:53 am

Perft(14) Weekly Status 2016-05-29

Symbolic has produced more than 46,100,000 perft(7) results so far, about 47.82% of the 96,400,068 needed.

Day count: 662
Estimated remaining day count: 722
Estimated total day count: 1,384

Average throughput: 69,637 results/day
Effective frequency: 178.03 GHz

Work units not yet started (490): 474-963

Sum of perft()s: 1,054,962,248,844,801,705
Sum of products: 10,182,675,144,596,924,219

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

The halfway point

Post by sje » Fri Jun 03, 2016 9:36 pm

The halfway point

The project has just passed the halfway point with respect to work units started (482), each having 100,000 positions from the unique(7) set.

In about two weeks, the number of completed work units should pass its halfway point.

mar
Posts: 2184
Joined: Fri Nov 26, 2010 1:00 pm
Location: Czech Republic
Full name: Martin Sedlak

Re: The halfway point

Post by mar » Fri Jun 03, 2016 10:14 pm

sje wrote:In about two weeks, the number of completed work units should pass its halfway point.
Congratulations! (hoping that you'll confirm Peter's count)

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Re: The halfway point

Post by sje » Fri Jun 03, 2016 11:01 pm

mar wrote:
sje wrote:In about two weeks, the number of completed work units should pass its halfway point.
Congratulations! (hoping that you'll confirm Peter's count)
So far, Symbolic with its 128 bit signatures has completed 69 work units (6,900,000 positions) in the 400-799 work unit range which was done by Ankan and his 64 bit signature program. All but about one in a half million results were confirmed with the mismatched results were all decided by TalkChess posters in Symoblic's favor.

As I understand it, Peter's program uses 64 bit signatures for the positions and a different set of 64 bit signatures for transposition table addressing. Will there be a match of his result and Symbolic's? If so, then we can all believe in the answer. Regardless, Symbolic will also produce proofs by giving the 20 ply one subtotals and the 400 ply two subtotals. Oh, and the 96,400,064 ply seven subtotals and multiplier factors.

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Re: The halfway point

Post by sje » Sat Jun 04, 2016 4:05 am

Ah, that's 96,400,068 ply seven subtotals.

----

Of the 69 work units in the 400-799 range completed, only eight have had mismatches. Which is pretty good.

----

On average, it takes about eleven days to process a work unit. The observed range is from under two days to just over 100 days depending on work unit difficulty and the machine in use.

----

On average, there are about 650,000 completed positions in 13 uncompleted work units being processed.

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Perft(14) Weekly Status 2016-06-05

Post by sje » Sun Jun 05, 2016 4:08 am

Perft(14) Weekly Status 2016-06-05

Symbolic has produced more than 47,100,000 perft(7) results so far, about 48.86% of the 96,400,068 needed.

Day count: 669
Estimated remaining day count: 700
Estimated total day count: 1,369

Average throughput: 70,404 results/day
Effective frequency: 184.69 GHz

Work units not yet started (480): 484-963

Sum of perft()s: 1,075,971,848,630,629,949
Sum of products: 10,675,158,870,493,097,745

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Perft(14) Weekly Status 2016-06-12

Post by sje » Sun Jun 12, 2016 5:00 am

Perft(14) Weekly Status 2016-06-12

Symbolic has produced more than 47,900,000 perft(7) results so far, about 49.69% of the 96,400,068 needed.

Day count: 676
Estimated remaining day count: 684
Estimated total day count: 1,360

Average throughput: 70,858 results/day
Effective frequency: 189.59 GHz

Work units not yet started (472): 492-963

Sum of perft()s: 1,092,555,411,890,157,343
Sum of products: 11,073,164,388,721,755,201

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Halfway point

Post by sje » Tue Jun 14, 2016 4:08 am

After 678 days of calculation, the Perft(14) project has passed the halfway point with 482 out of 964 100,000 position work units completed PLUS one straggler work unit with only 68 positions.

----

An assortment of tough positions, the largest one from each completed work unit: https://dl.dropboxusercontent.com/u/316 ... ar/Largest

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Perft(14) Weekly Status 2016-06-19

Post by sje » Sun Jun 19, 2016 4:05 am

Perft(14) Weekly Status 2016-06-19

Symbolic has produced more than 48,500,000 perft(7) results so far, about 50.31% of the 96,400,068 needed.

Day count: 683
Estimated remaining day count: 675
Estimated total day count: 1,358

Average throughput: 71,010 results/day
Effective frequency: 194.71 GHz

Work units not yet started (466): 498-963

Sum of perft()s: 1,109,933,223,233,994,493
Sum of products: 11,490,231,860,973,846,801

User avatar
sje
Posts: 4675
Joined: Mon Mar 13, 2006 6:43 pm

Perft(14) Weekly Status 2016-06-26

Post by sje » Sun Jun 26, 2016 4:04 am

Perft(14) Weekly Status 2016-06-26

Symbolic has produced more than 49,300,000 perft(7) results so far, about 51.14% of the 96,400,068 needed.

Day count: 690
Estimated remaining day count: 659
Estimated total day count: 1,349

Average throughput: 71,449 results/day
Effective frequency: 202.88 GHz

Work units not yet started (458): 506-963

Sum of perft()s: 1,135,571,352,229,705,732
Sum of products: 12,094,604,291,665,471,869

Post Reply