7 8 or more pieces tablebase generator

[pPerez]

I'm new member to this site, so I would like to say a big hello

I am an endgame enthusiast. When I first knew of Nalimov tablebases I was so impressed that a tool could play chess perfectly, that I decided to develop my own tablebase generator.

Depending on the position caracteristics, it can resolve positions with up to 7 or 8 pieces. Even more for pawn endgames.

In my real job, I work as a computer engineer for Airbus, in Toulouse, France. It took me 4 years to develop this tool in my leisure time.

If you are interested, I invite you to visit my website

http://www.mtu-media.com/finalgen

Try it, it's easy to use, and free!

Pedro Pérez Romero

[Kirill Kryukov]

Very interesting work, thanks a lot for sharing! Also, welcome to this forum!

[jshriver]

Best of luck to your endeavor. Look forward to following this.

[Arpad Rusz]

A very nice tool! Thanks!

[Arpad Rusz]

Some discussion about FinalGen on the Talkchess forum:
http://talkchess.com/forum/viewtopic.ph ... w=&start=0

[pPerez]

.

[JohnnyBob]

FinalGen
http://www.mtu-media.com/finalgen/home_ing.php
FinalGen is a chess endgame tablebase generator for Windows.

There is also a long thread on the Rybka forum about FinalGen:
http://rybkaforum.net/cgi-bin/rybkaforu ... ?tid=24565

It discusses, among other things, the flags it raises with various anti-virus software. I believe the general final concensus there is that it probably has no viruses, but needs to do something to be more compatible with anti-virus software. For example I can't even install it with avast as my anti-virus, which flags and quarantines most if not all of the dll files during the install process.

I tried it that way, with avast. Perhaps I could create exclusions with avast, but I didn't try that (yet). Note that I only installed it up to the point where it asked me if I wanted to "run" the software, which I declined. Apparently it wasn't completely installed at that point, because when I attempted to uninstall it later there was an error message about not being able to read the uninstall.ini file, or some such. Then I had to do a manual uninstall which was a bit complicated, and I'm not absolutely sure I got everything.

It's very interesting software(!) and I hope these two problems (conflicts with anti-virus software, and proper operation of the installer/uninstaller) can be fixed soon, then I will consider investigating it further. I believe it may be very useful software, representing a significant advance in endgame analysis and solving. My thanks to the author for his considerable efforts.

[JohnnyBob]

As an update...
I switched to a different anti-virus and FinalGen installed OK, but it still wouldn't run because it requires SSE2. So there's no point to installing it on older pre/non-SSE2 processor computers. [It then uninstalled cleanly this time.] I have another computer available that probably has SSE2(?) but am not sure when I can find the free time to test it there...

[pPerez]

Hi

The Version 1.1 of Finalgen has been released, with the following enhancements:
1- A new status "Unsolved" was introduced to differentiate unsolved positions from non-generated ones. Next moves of unsolved positions are available.
2- New Copy/Paste FEN buttons. Although no GUI modifications were initially planned, many people asked for those buttons, and this is the reason why they were implemented.


http://www.mtu-media.com/finalgen

Pedro

[JohnnyBob]

Thanks for the heads up. I installed the new 1.1 version of FinalGen on a Windows 7 x64 computer with 4 GB RAM and tested it with a few simple quick positions. It seemed to work fine.

That gave me courage to try a 9 piece position, total tablebase size 307GB. I tried first with an external USB 2.0 drive but it wanted to take like 40-50 hours which was too long. I aborted that, and switched to my computer's internal hard drive and it estimated about 11 hours for the job. So I allowed it to proceed and it seemed to be working OK.

Unfortunately after running FinalGen for 8+ hours and finishing 72% of the job, the computer froze when I pressed the FN key on the keyboard, with a (Not Responding) error message in the titlebar of FinalGen. Pressing the FN key is not an operation in FinalGen. I pressed the FN key for a different unrelated purpose. I was just playing around. I have no idea why pressing the FN key would cause the computer to freeze, except that this particular computer has a history of freezing at random occasionally. I wasn't multitasking any other software, just running FinalGen.

So I rebooted computer (required to recover) and tried to resume the tablebase generation, but it said an unexpected error requires closing FinalGen. So resuming the tablebase generation after such an interruption (computer freeze) is apparently not possible. Thus the 265GB of data which it had generated is apparently worthless. I didn't have the heart to start over, so have put FinalGen on the back burner for now. I need that computer for other tasks. I might try it again later when I find more free time.

It's very interesting software!

[Luigi335]

Thank you, Pedro, for the version 1.1 of FinalGen, that perfectly works on my PC

I have tried an exploit at limit of impossible: the 11 men endgame of the Spassky - Fischer, first of 1972 Rejkiavik Match, after 37.Ba3
Of course time (240 hours) and memory (4 TB) are, at the moment, prohibitive, but I am sure that in the next years, for the future CPUs and hard disks, this will be no longer a limitation, and such a position will be solved in only one night!

Thank you again!

[JohnnyBob]

I installed FinalGen v1.1 on a different computer with Windows XP Pro SP3, 2GB RAM, Intel Core2 Duo T9300 @2.5GHz. It installed OK and I was able to generate 52GB 548 MB tablebases for a 7-piece position in 1h25min19s. But I'm still a newbie with FinalGen and have a few questions please.........

I used this FEN starting position:
8/8/8/4k3/4p3/2K3P1/2N2P2/2n5 w - -

With the above FEN contained in windows clipboard, I clicked the "Paste FEN" button in FinalGen. That produced a popup box with an error message saying "FinalGen: Could not paste the position. OK". However there seems to be no real error, because after clicking the OK button, it was correctly pasted nevertheless. So apparently that error message (cause unknown) can be ignored as of no consequence.

After the tablebases were generated...

I can doubleclick one of the moves in the list to move forwards, which sets up that new position on the chessboard.
How do I move backwards to the prior position on the chessboard

After playing around with analysis and variations for awhile, I have lost the original chessboard position that the tablebases were generated for, and the analysis area below the chessboard is empty.
So how do I get back to the original chessboard position

If I exit then restart FinalGen:
How do I know which positions tablebases were previously generated for
How do I bring up one of those positions on the chessboard

Thank you for any help!

[pPerez]

Hello JohnnyBob

How do I move backwards to the prior position on the chessboard?
Sometimes, when I analyse a position I also look for a "back" button, but I don't find it
In those cases, I move backwards manually

So how do I get back to the original chessboard position?
The button "Initial position" retrieves on the chessboard the last position for which a generation process was launched

How do I know which positions tablebases were previously generated for ?
How do I bring up one of those positions on the chessboard?
In fact, when tablebases are generated, millions or billions of positions are generated and stored in the working folder. For now, It is not possible to group them into categories, like "all KNPKBP are in the database".
However, FinalGen could store the initial positions that were generated, so it could list or retrieve them, but this feature has not been implemented. Maybe in a future version.

Regarding the problem you had on your computer, you can re-launch a generation process after a crash, but you should re-generate the tablebases from scratch (not resume the old process, otherwise you can get an "unexpected error"). The data that are already in the tablebases will not be generated again, so you don't need to wait during 11 hours again.

PS: If you have other questions, you can contact me at this address finalgen@mtu-media.com. I am not sure I can answer on forums

Regards
Pedro Pérez Romero

[byakuugan]

I am back from the World Open where I obtained a GM norm, so I have a lot of free time right now. I've been studying some endgames with FinalGen (namely the drawing fortresses in BPP vs. RPP, and NP vs. BPP) and I wanted to ask how hard it would be to design a program for fairy endgames with one piece per side and the rest pawns. It would be interesting to study which pieces are better at holding fortresses than others, and also the long wins that come along with breaking the semi-fortresses. Namely I am only interested in leapers (camel, giraffe, zebra, dabbaba, alfil, ferz, wazir, etc.)

[JohnnyBob]

I've been using FinalGen (version 1.3) more and more and am getting used to it. It's a regular part of my chess arsenal now. It seems to be stable software, no crashes.

I've used it to solve endgames with 9-10 pieces but of course 7-8 piece positions are usually more practical. Sometimes it only takes an hour or less, sometimes a day or more. External USB drives come in handy because some positions can require 1-2TB or more of storage space, but I can run most 7-8 piece positions just with my 640 MB internal hard drive.

Nice thing is that it coexists happily with other chess software such as a chess GUI and chess engine running on the same machine, no problem, and only uses 1 processor core for it's calculations. No special hardware required except a more modern processor is needed with the SSE2 function.

Interested persons should be sure to check out the new forum at
http://www.mtu-media.com/finalgen/home_ing.php
It has some nice custom features and the FinalGen author is very helpful. There is already some useful info posted there about the detailed nature of this software and its limitations. Stop by sometime and say hello!