Attack the 8-piece tablebases

Discussion of chess software programming and technical issues.

Moderator: Ras

natopigro
Posts: 5
Joined: Tue Sep 21, 2021 5:44 pm
Full name: Giorgio Ricca

Attack the 8-piece tablebases

Post by natopigro »

Hi all,
I am a data scientist and a chess player (2000 elo).
Let say I have acces to a super computer, it's in the top N where N is not so big. I have the possibility to use it for, let say, "personal" purposes. I was wondering how to exploit it (legally) and I was attracted by the 8-piece tablebases problem.

I know that at the moment the problem is so big that basically it is not feasible to solve it, but I could try to give my contribution solving a subset of positions. Also for my company, a valid results can be seen as good publicity.

Is there someone which is interested in collaborating? Maybe someone with any experience in that field?
Where can I start from? I don't want to reinvent the wheel, maybe there is some code in Python i could try to reuse?

I am open to questions from anyone interested.

Bye!
User avatar
phhnguyen
Posts: 1524
Joined: Wed Apr 21, 2010 4:58 am
Location: Australia
Full name: Nguyen Hong Pham

Re: Attack the 8-piece tablebases

Post by phhnguyen »

natopigro wrote: Thu Sep 23, 2021 7:25 pm Hi all,
I am a data scientist and a chess player (2000 elo).
Let say I have acces to a super computer, it's in the top N where N is not so big. I have the possibility to use it for, let say, "personal" purposes. I was wondering how to exploit it (legally) and I was attracted by the 8-piece tablebases problem.

I know that at the moment the problem is so big that basically it is not feasible to solve it, but I could try to give my contribution solving a subset of positions. Also for my company, a valid results can be seen as good publicity.

Is there someone which is interested in collaborating? Maybe someone with any experience in that field?
Where can I start from? I don't want to reinvent the wheel, maybe there is some code in Python i could try to reuse?

I am open to questions from anyone interested.

Bye!
If you want a serious discussion, you should post or require Moderators to move this post into the sub-forum Programming and Technical Discussions since almost all our developers discuss in that sub-forum that but not here. I think many people will be interested and helpful.

Secondly, you may read some text/posts first, including posts of these forums about Syzygy from CPW (link is below). Many forums' posts have mentioned already about 8 men.

https://www.chessprogramming.org/Syzygy_Bases

Good luck and have fun with your project :)
https://banksiagui.com
The most features chess GUI, based on opensource Banksia - the chess tournament manager
dangi12012
Posts: 1062
Joined: Tue Apr 28, 2020 10:03 pm
Full name: Daniel Infuehr

Re: Attack the 8-piece tablebases

Post by dangi12012 »

natopigro wrote: Thu Sep 23, 2021 7:25 pm Hi all,
I am a data scientist and a chess player (2000 elo).
Let say I have acces to a super computer, it's in the top N where N is not so big. I have the possibility to use it for, let say, "personal" purposes. I was wondering how to exploit it (legally) and I was attracted by the 8-piece tablebases problem.

I know that at the moment the problem is so big that basically it is not feasible to solve it, but I could try to give my contribution solving a subset of positions. Also for my company, a valid results can be seen as good publicity.

Is there someone which is interested in collaborating? Maybe someone with any experience in that field?
Where can I start from? I don't want to reinvent the wheel, maybe there is some code in Python i could try to reuse?

I am open to questions from anyone interested.

Bye!
Hello I can see you wrote the same thing on reddit 2 months before and I suggested you pick up the Tablebase problem. Good thing you picked up the idea!
You should pm the contributors of this repo:
https://github.com/syzygy1/tb


8 Man TB is totally possible in 2021 (but its a lot of work!). Hardware made some big big leaps especially gpu -SSD and interconnect stuff. For example 15GB sequential read was DDR2 speed once. But now you can get this speed with NVME Raid Array. Gpu memories increased from 786MB to 24-48GB so that should help too..
Worlds-fastest-Bitboard-Chess-Movegenerator
Daniel Inführ - Software Developer
R. Tomasi
Posts: 307
Joined: Wed Sep 01, 2021 4:08 pm
Location: Germany
Full name: Roland Tomasi

Re: Attack the 8-piece tablebases

Post by R. Tomasi »

dangi12012 wrote: Fri Sep 24, 2021 5:11 pm 8 Man TB is totally possible in 2021 (but its a lot of work!). Hardware made some big big leaps especially gpu -SSD and interconnect stuff. For example 15GB sequential read was DDR2 speed once. But now you can get this speed with NVME Raid Array. Gpu memories increased from 786MB to 24-48GB so that should help too..
From what the OP said, I would think that he isn't intending to do this on normal hardware :wink:
dangi12012
Posts: 1062
Joined: Tue Apr 28, 2020 10:03 pm
Full name: Daniel Infuehr

Re: Attack the 8-piece tablebases

Post by dangi12012 »

R. Tomasi wrote: Mon Sep 27, 2021 12:02 pm
dangi12012 wrote: Fri Sep 24, 2021 5:11 pm 8 Man TB is totally possible in 2021 (but its a lot of work!). Hardware made some big big leaps especially gpu -SSD and interconnect stuff. For example 15GB sequential read was DDR2 speed once. But now you can get this speed with NVME Raid Array. Gpu memories increased from 786MB to 24-48GB so that should help too..
From what the OP said, I would think that he isn't intending to do this on normal hardware :wink:
What do you think? Supercomputers dont have SSDs or GPUs? Think again. https://www.top500.org/lists/top500/2021/06/
Many use clusters of Nvidia A100 with Mellanox interconnect etc. Its perfect for chess. Someone with access needs to put in the work together with tablebase creators. 8 Man Tablebase is totally in reach by a 2021 supercomputer.
Worlds-fastest-Bitboard-Chess-Movegenerator
Daniel Inführ - Software Developer
natopigro
Posts: 5
Joined: Tue Sep 21, 2021 5:44 pm
Full name: Giorgio Ricca

Re: Attack the 8-piece tablebases

Post by natopigro »

Hi all, thank you for your suggestion.
I've already written to syzygy_tb@yahoo.com but he didn't replyed up to now.

For sure, to the moderators, please feel free to move my post in the proper section.

@dangi12012 yes, it was me on reddit, thank you for your suggestion :D

The HPC I have the possibility to use definetely has the aforementioned features. I would just need a code to run :mrgreen:
User avatar
phhnguyen
Posts: 1524
Joined: Wed Apr 21, 2010 4:58 am
Location: Australia
Full name: Nguyen Hong Pham

Re: Attack the 8-piece tablebases

Post by phhnguyen »

If you really start, I guess Ronald de Man and/or other people will jump in to help you. Similar to the way they helped Bojun Guo to generate 7 men.

http://www.talkchess.com/forum3/viewtopic.php?t=66797
https://banksiagui.com
The most features chess GUI, based on opensource Banksia - the chess tournament manager
pepechuy
Posts: 226
Joined: Fri Oct 22, 2021 4:22 am
Full name: José García Ruvalcaba

Re: Attack the 8-piece tablebases

Post by pepechuy »

scchess
Posts: 51
Joined: Mon Jan 04, 2021 4:12 pm
Full name: scchess

Re: Attack the 8-piece tablebases

Post by scchess »

Solving the 8-man TB is too much, a good starting point should just focus on certain well known 8-man configurations such as R+P vs R+P.
Chessnut1071
Posts: 313
Joined: Tue Aug 03, 2021 2:41 pm
Full name: Bill Beame

Re: Attack the 8-piece tablebases

Post by Chessnut1071 »

natopigro wrote: Thu Sep 23, 2021 7:25 pm Hi all,
I am a data scientist and a chess player (2000 elo).
Let say I have acces to a super computer, it's in the top N where N is not so big. I have the possibility to use it for, let say, "personal" purposes. I was wondering how to exploit it (legally) and I was attracted by the 8-piece tablebases problem.

I know that at the moment the problem is so big that basically it is not feasible to solve it, but I could try to give my contribution solving a subset of positions. Also for my company, a valid results can be seen as good publicity.

Is there someone which is interested in collaborating? Maybe someone with any experience in that field?
Where can I start from? I don't want to reinvent the wheel, maybe there is some code in Python i could try to reuse?

I am open to questions from anyone interested.

Bye!
I understand what you're doing. If your boss finds out you could be in serious trouble. It's not worth a job. Strongly advise you to use your own time and get access to a computer with that capability at your own expense. I know people who have been there, done that , have the t-shirt and are currently unemployed.