after 2020 right there not technology
Progress of 8 piece tablebase?
Sort:
8-piece in DTM is impractical. It should be generated in syzygy format.
DTM format is bulky and cumbersome.
8-piece is more easy to handle and maintain if more compact. Atleast in the retention aspect of the project. It has to be stored somewhere. 1-2 PB should be feasible today.
Apr 11, 2020
0
#6
DamonevicSmithlov wrote:
If one side has a king, rook, & pawn & the other side has king, rook, & 2 pawns is that considered 7 piece tablebase? Or do they have to be seperate pieces, like what I just described is only 3 piece tablebase (king, rook, pawn)?
The number refers to the total number of units on the board. The computer power required to generate an 8-piece tablebase is absolutely massive – an exponential increase compared to 7-piece. Even if someone were to spend the millions on the project, it would take decades to finish, if ever. That's my guess, anyway.
Check out my introduction to the subject: Adventures with endgame tablebases.
Short of new memory technology, perhaps holographic for the average desktop computer, it would be impractical to tie up a petabyte of cloud storage for an 8 man tablebase.
There seems, to the average person that the cloud has unlimited capacty, but a petabyte would be enough storage to be an monitary issue: who would pay the cost to host that?
I do think eventually, someday, with advances in memory technology, that it would eventually be cost effective.
Years ago, people marveled at a company hosting 5 gigabytes of data *online, but over the years it crept ip to 10 gigs, then 20 gigs, but compnies are offering to host 50 and even 100 gigs or more as part of some sort of tiered product subscription.
A petabyte needed to host an 8 man tablebase?... eventually.
*Usually thrown in as part of a subscription for another product.
Apr 16, 2020
0
#8
The storage requirements for 8-man tablebases is enormous, but I'd point out the complete set of 8-man tablebases does not have to be built all at once. The positions without pawns are independent of each other, and they have to be built first anyway. So build them one at a time, knowing they only require the 7-man tablebases for a complete resolution. Granted chessplayers would find tablebases with pawns to have more practical value, but the more statistically-minded folks would be interested in how many moves are required for some of these checkmates.
There certainly will be benefits of using of 8-men for engines.
This position arose recently in this game.
... to handle 1PB ( = 1000 TB ) for the whole 8-pieces endgame tablebase is realy peanuts for him https://www.flickr.com/photos/olcf/sets/72157697679727475/
1PB is also no problem for HAWK in Stuttgart ( place 5 in the world )
worldrecord-evolution for longest mate composition ( until mate in 553 moves ) :
https://www.youtube.com/watch?v=bqnq580ozAI
... or perhaps you like 400 hard worldwide unsolved puzzles :
https://gameknot.com/chess-puzzles.pl?s=0&a=1&u=chesscode&m=0&s=9
or try to find this forced mate in over 1000 moves if you want :
https://www.youtube.com/watch?v=zsjLqzvbkWA
kindly regards
Lutz Neweklowsky
Karlsruhe ( Germany )
Caesar49bc wrote:
Short of new memory technology, perhaps holographic for the average desktop computer, it would be impractical to tie up a petabyte of cloud storage for an 8 man tablebase.
There seems, to the average person that the cloud has unlimited capacty, but a petabyte would be enough storage to be an monitary issue: who would pay the cost to host that?
I do think eventually, someday, with advances in memory technology, that it would eventually be cost effective.
Years ago, people marveled at a company hosting 5 gigabytes of data *online, but over the years it crept ip to 10 gigs, then 20 gigs, but compnies are offering to host 50 and even 100 gigs or more as part of some sort of tiered product subscription.
A petabyte needed to host an 8 man tablebase?... eventually.
*Usually thrown in as part of a subscription for another product.
This is silly. A petabyte is only 1024TB. I have around 1.5TB in the cloud, and I have unlimited cloud storage with my EDU e-mail. I can have a petabyte up there if I want! I'm one single person. Clearly it is not that hard to support 1000TB for an incredibly useful service. But yes, hosting on a server would surely be an issue. But honestly, we aren't even that far away from being able to store on physical drives. Costs like $200 for a 12TB HDD. Could at least store the most common positions pretty easily right now, and 10 years down the line we'd def be able to handle a petabyte.
I know this is an older thread but, before I even looked into 8 piece tablebase, I doubted that without some revolutionary tech - maybe like million-q-bit quantum computers with awesome programmers , that we wouldn't see 8 piece chess solved in our lifetime. Just because it's the number 8. Numbers seem to have their own characteristics and quirks. 8 seems to oftentimes open doors to the completely limitless. Hence the chessboard itself is an 8 by 8 system.
Looking at the nature of numbers; adding one more to a seven-system , and getting to the number 8 - seems to always throw a "monkey wrench" into things, and kick patterns off into repeating systems, and is likely why the infinity symbol is an 8 on it's side. When a music scale repeats it at the eighth step - (called an octave) I mention this and some other examples, in a quick-music talk video, and mention the similarities between chess and music, if anyone is interested it's here: https://www.youtube.com/watch?v=YMvAcGrFeA4
One of the commenters above mentioned that they "estimate" a petabyte? Is that 1000 terrabytes? I mean, not even a for-sure thing. and that sound completely bonkers to solve that using the computers we have now, even as amazing as they seem to us now. Cheers 
DiCello wrote:
I know this is an older thread but, before I even looked into 8 piece tablebase, I doubted that without some revolutionary tech - maybe like million-q-bit quantum computers with awesome programmers , that we wouldn't see 8 piece chess solved in our lifetime. Just because it's the number 8. Numbers seem to have their own characteristics and quirks. 8 seems to oftentimes open doors to the completely limitless. Hence the chessboard itself is an 8 by 8 system.
Looking at the nature of numbers; adding one more to a seven-system , and getting to the number 8 - seems to always throw a "monkey wrench" into things, and kick patterns off into repeating systems, and is likely why the infinity symbol is an 8 on it's side. When a music scale repeats it at the eighth step - (called an octave) I mention this and some other examples, in a quick-music talk video, and mention the similarities between chess and music, if anyone is interested it's here: https://www.youtube.com/watch?v=YMvAcGrFeA4
One of the commenters above mentioned that they "estimate" a petabyte? Is that 1000 terrabytes? I mean, not even a for-sure thing. and that sound completely bonkers to solve that using the computers we have now, even as amazing as they seem to us now. Cheers
A petabyte is 1024 terabytes. We will near-definitely be able to get 8-men tablebase in our lifetimes, perhaps even in the next few decades. The storage requirement is not an issue at all as Syzygy estimates roughly 2PB. High-end servers can manage 2PB if needed. Solutions can also just be stored piece by piece, so it's a non-issue. The problem is the computational resources including RAM that will be required of the supercomputer, and furthermore, actually securing that supercomputer for solving chess positions. We could probably do this even today if we had access to a Top 10 supercomputer and sufficient allotted time.
With every increase in 1-man, the resources required drastically goes up. I imagine 9-man and beyond will be impossible to complete within our lifetimes.
8-men tablebase in Syzygy format requires 2 petabyte.
Anyone who knows the latest news about the 8-man project. Have the Russians put it on hold or will it probably be feasible next couple of years.
Most of the "interesting" pawnless 8-man endings (i.e. the ones in which the material difference is not "too large") have already been solved, using the DTC metric, courtesy of Marc Bourzutschky. I believe Marc has also solved all of the 8-man endings he refers to as the "OP-1 set" (i.e. those containing one pair of opposing pawns on the same file, preventing the possibility of promotion without at least one capture being made). Here are the links:
Oh, and a win in 584 moves (again, using the DTC metric) :
@17
Two dark square bishops...
Yes, one of Bourzutschky's more unfortunate discoveries was that many of the deeper 8-man endings require a promoted bishop, and will therefore never occur in serious play. Here is the longest win (in terms of DTC) found by Bourzutschky in an 8-man ending with no promoted pieces:
Dec 10, 2022
0
#19
Considering the incredible increasing rate of computational power and storage for computers, at all levels, over the last 50 years, that somehow an 8 man table base would be some sort of difficult task to either complete or store soon is a bit odd. By 2030 this will be seen as a rather easy task. I actually dread quantum computing getting involved in chess, an engine using a rather powerful quantum computer will very likely, this century, be able to tell you during the middle game that you have lost, won or drawn with correct play.
First 6 piece tablebase published was Nalimov tablebase, sometime in 2005. It takes up 1,2 TB.
First 7 piece tablebase published was Lomonosov tablebase, in August 2012. It takes up 140 TB.
How much would a 8 piece tablebase take up?
Making the 7 piece tablebase after 6 piece tablebase took about 7 years, 2005 to 2012.
It is now 6 years and 2 months since 7 piece tablebase was completed, August 2012 to October 2018.
Is anybody known to be computing a 8 piece tablebase?
How much is left to compute of the 8 piece tablebase?