tablebases

A simple rule to evaluate the  successive  size of chess tablebases  each time it is increased by one piece is to multiply the size by 64 each time an extra piece is added.This rule is especially right if there is no compression in any of the lesser piece tablebases used to make the calculation.i thought  that the nalimov table was only 30 gigs for all 6 piece?

I'm not sure where the assumption of multiplying the size of an n-man tablebase by 64 to get the tablebase size for n+1 men comes from, but it surely doesn't seem sound. For instance, 4-man tablebases are almost 500 times larger than 3-man tablebases (Nalimov format).

3-man - ~63 KiB    (fits easily onto a single floppy)

4-man - ~30 MiB    (needs a few dozens of floppies, or a single CD)

5-man - ~7.05 GiB    (needs a dual-layer DVD)

6-man - ~1.2 TiB    (needs a 1.5TB HDD)

7-man - hundreds of TiB    (needs a small data center)

8-man - unknown, but probably on the order of tens of PiB    (would need a large data center)

9-man - ?

...

As for solving chess via tablebases, it will never happen. If every atom of matter comprising the Earth could hold a bit of data, it probably would not be enough to store full 16-man tablebases. As for complete 32-man tablebases, the matter in all of our galaxy, or even the Universe, might not suffice. The tablebases that have been created thus far, and likely those ever to be created, are a mere tip of an iceberg of sheerly unfathomable proportions.

 sorry my math was off by a factor of 12!  I forgot to account for the fact that when u ad a piece to a data base since u have choice of 12 pieces, the data base would expand 64 squares X 12 pieces  

You're quite late with the apology? :)

jaaas wrote:

I'm not sure where the assumption of multiplying the size of an n-man tablebase by 64 to get the tablebase size for n+1 men comes from, but it surely doesn't seem sound. For instance, 4-man tablebases are almost 500 times larger than 3-man tablebases (Nalimov format).

3-man - ~63 KiB    (fits easily onto a single floppy)

4-man - ~30 MiB    (needs a few dozens of floppies, or a single CD)

5-man - ~7.05 GiB    (needs a dual-layer DVD)

6-man - ~1.2 TiB    (needs a 1.5TB HDD)

7-man - hundreds of TiB    (needs a small data center)

8-man - unknown, but probably on the order of tens of PiB    (would need a large data center)

9-man - ?

...

As for solving chess via tablebases, it will never happen. If every atom of matter comprising the Earth could hold a bit of data, it probably would not be enough to store full 16-man tablebases. As for complete 32-man tablebases, the matter in all of our galaxy, or even the Universe, might not suffice. The tablebases that have been created thus far, and likely those ever to be created, are a mere tip of an iceberg of sheerly unfathomable proportions.

xxxxxxx

This is stupid wrong

Using current computer storage medium as the basis for limits is flawed. It's now possible to store hundred of terrabytes onto a disc which you could balance on a finger (http://www.gizmodo.co.uk/2016/02/optical-data-storage-squeezes-360tb-on-to-a-quartz-disc-forever/). When this mechanism is refined and becomes mainstream I'm sure we'll start seeing new levels of TB depth. 

The most promising way to mathematically solve chess is to "compress" these table bases into an algorithm instead of a database which is very small.  If that ever happens eventually 32 piece solved chess could be a reality.  Solving backwards from the winning end positions seems much more promising than starting with the opening.  Although such an algorithm is very difficult to find.  Even if generalized board size chess is EXP hard problem, 8x8 can have a reasonable compact perfect algorithm theoretically speaking.

Um... it's been VERY long since any activity has occured on this page but I think 7-piece endgames have now been solved.