When will 8 piece tablebase be ready?
Moderators: hgm, Rebel, chrisw
-
- Posts: 70
- Joined: Thu Sep 19, 2019 5:55 am
- Full name: Andy!
-
- Posts: 2821
- Joined: Fri Sep 25, 2015 9:38 pm
- Location: Sortland, Norway
Re: When will 8 piece tablebase be ready?
Much appreciated. Was not aware of thisthe_real_greco wrote: ↑Sun May 30, 2021 1:47 am http://arves.org/arves/index.php/en/lat ... plorations
-
- Posts: 12038
- Joined: Mon Jul 07, 2008 10:50 pm
Re: When will 8 piece tablebase be ready?
Great stuff.the_real_greco wrote: ↑Sun May 30, 2021 1:47 am http://arves.org/arves/index.php/en/lat ... plorations
Anyone know how long to complete ?
-
- Posts: 65
- Joined: Sun Aug 19, 2018 10:57 pm
- Full name: Nickolas Reynolds
Re: When will 8-man EGTB be ready?
You need more storage to make the tablebases than you do to store the finished product.Ajedrecista wrote: ↑Sat May 29, 2021 8:36 pmIf a ratio of 0.05 bytes/NULP is assumed, then the whole set of 8-man EGTB would be circa 1.7 PiB in size, way below 10 to 20 PiB mentioned above if my fast math is not wrong.
-
- Posts: 1968
- Joined: Wed Jul 13, 2011 9:04 pm
- Location: Madrid, Spain.
Re: When will 8-man EGTB be ready?
Hello Nickolas:
I returned to NULP site and extracted some data on 7v1, 6v2, 5v3 and 4v4:
Then, if we assume that intermediate and final sizes are more less proportional to NULP, the logical strategy looking for minimum storage during the generation task would be starting with the biggest subset of EGTB just to get rid of the bigger intermediate results earlier: firstly 5v3, secondly 4v4, then 6v2 and finally 7v1. I know, easier to say than doing it. Any insight is welcome.
Regards from Spain.
Ajedrecista.
Sure, I was only thinking on the final release. Good point. I do not know how much storage will be needed for the intermediate results in comparison with the final results, by the way. However, I think you do not need to save all the intermediate results, do you? If you start with 7v1 as a test, once you finish them you can (safely?) delete all but the final results because 7v1 will not be useful for 6v2, 5v3 and 4v4 generations. Then you go with 6v2 for example and once you are done, then delete the intermediate results of 6v2 and so on.
I returned to NULP site and extracted some data on 7v1, 6v2, 5v3 and 4v4:
Code: Select all
Type Number of endgames NULP
------------------------------------------------------
7v1 210 1,292,210,487,827,262
6v2 630 7,355,226,093,580,581
5v3 1,050 17,795,943,153,932,922
4v4 630 11,732,927,142,407,480
------------------------------------------------------
TOTAL 2,520 38,176,306,877,748,245
Regards from Spain.
Ajedrecista.