The concept of SMIRF tries to cover a lot of different 8x8 and 10x8 chess variants compatibly integrating the traditional chess game, Fischer Random elements and Capablanca's 10x8 extended piece set.

It's main application CRC (Capablanca Random Chess) has been created to provide a new testing field for chess programming, thus it has not been intended mainly to be a game between human players. Nevertheless it already seems to be attractive to some fans.

SMIRF also is able to play Gothic Chess for internal purposes only (testings, competitions). Maybe there sometimes might be an agreement to generally open its GC awareness to every user.



SMIRF is able to read and store its games in PGN format (files and clipboard). It also supports the possibility to comment all or selected moves manually. You can watch the current PV lines in a special window, which also is enabled to provide clipboard copies. When started the program will automatically restore the last shut down situation without any needed further steps.

SMIRF is intended to become Shareware. Actually only a Beta version exists.

SMIRF Beta download possibility (free until end of September 2005)
(Hint for beta testers: secure and delete existing SMIRF INI files before you install the current version.)

Regards, Reinhard.

Hello Reinhard,

And welcome to the board! I wish a warm welcome to you and to S.M.I.R.F. I look forward to playing some games with your program on GothicChessLive.com some time, I really enjoyed some of our previous encounters.

Did you save any of these games? Would you like to post them in the "Great Games" section, then I can annotate them?

See you later!

--Ed

Hello Ed,

since SMIRF is still unready, played games have been regarded as experimental results. So they did not survive for more days. Maybe the actually running friendly games are better candidates to be commented later. As already noticed, there exist different evaluation philosophies, thus the games will have a lot of controversly to be interpreted situations.

Regards, Reinhard.

Hi Reinhard,

I view the commentary as interesting and valuable. I do not think there is a "right way" or "wrong way" to design such an intricate piece of software. I was just curious as to how certain evaulations were arrived at. I let Vortex search for a while since you were reporting a deeper search depth. If you would like the commentary to stop, I will abide by your wishes, or is your remark more to Andreas, who was commenting as we played?

Hi Ed,

commenting is welcomed. Previous posting has been intended as an excuse for me not to be able to supply you with more valuable game notations. I regard your comments, Ed, as some being related to an experienced chess player. Neither the evaluation routines in SMIRF nor my interpretations are made from such a viewpoint. I have paused for very long from active playing chess to get sufficiently far from playing practice, because I am convinced, that else this would lead to an attempt to simulate a playing self. But this is something, I want to avoid in any case. Thus Smirf's evaluation might be constructed very distinct from an experienced player's view. Contradictingly to this approach it is very interesting to learn how a more traditional view would comment and evaluate positions compared to Smirf's evaluations.

Reinhard.

Nice looking GUI Reinhard!

Good, if people would like it! Note, that Chancellor and Archbishop have different symbols at SMIRF than in Gothic Vortex. I hope by that to match better their gaits and to separate them clearer from traditional pieces.

I am still on search for a sponsor to support this project. Maybe the whole application could then be improved and moved to a Mac as an example. Also multiprocessing abilities still have to be developed.

Reinhard.

If SMIRF is going to be shareware, or even freeware (?), is there any chance that some of you programmers out there could recompile it for a DOS platform? It would naturally have to be scaled down, and users would have to be satisfied with fairly shallow searches.

Hi Beta-Zero,

SMIRF is intended to become Shareware. The TMCI protocol used by it is free. But actually there are no other programmers interested to use it. I do not understand the request for scaling it down for DOS, so please explain your needs. But there is an idea to write an engine variant using a protocol like UCI, that would imply a possible usage of the engine at a command prompt, but it still would need 32 Bit windows.

Reinhard.

If I go to the local computer store, Windows XP on a CDROM costs $299, a bit expensive for someone like me.

My Windows 98 computer crashed big time a couple months ago, leaving me pretty much at the mercy of DOS based applications. (Come to think of it, I haven't the slightest idea where the 98 system disk went, but I am beginning to THINK my Dad lent it to somebody 5 years ago, and now I am out of luck.)

Now, I'm not a Windows programmer, and I am certainly not a DOS programmer.

So, unless the source code to SMIRF is generic 680x0 assembly language, with tons of comments all over the place, I wouldn't be able to use it anyway. I was hoping that somebody out there could recompile the SMIRF code so it executes on DOS.

Vortex looks like a good program too, but I don't know where I am going to find a PC running Windows to try it out on.

The Mac, of course, is a good computer but my finances prohibit me from getting one.

Hi Beta_Zero,

SMIRF is a 32-Bit Windows program. It needs Windows 95 or newer. I do not intend to scale it down to the ancient 16-Bit world.

Reinhard.

There is a new beta available, free and unrestrictedly usable (sorry, no GothicChess yet) including October 2005. There obviously have been some difficulties with prior betas to have SMIRF think unrestrictedly.

Download the new version from http://www.chessbox.de/Compu/schachsmirf_e.html , but do not forget to delete prior INI-files of SMIRF and uninstall it before the new one (or if need be enter later manually following testing keys):

Regards, Reinhard.

Well, I have erased a bug, which enabled bad values to be accumulated in the cache. Thus there is a new beta 1.19.

Seeing how the endgame databases really did save the game for Vortex last time, I am focussing on extending this capability further. In an experimental version of Vortex, I am not terminating a search if there are 7 or 6 pieces on the board. Instead, I do a very fast 3-ply search to see if the endgame databases can be hit.

My first implementation of this check extension code was bad. The game tree in my test positions was 5 times as large! A 12 ply search requiring 30 seconds to complete was now taking 6 minutes.

But, it was hitting the 5-piece database 3 times as often.

I am tuning the conditions for this check extension. If one side has 2 or more pawns in the 7-piece position, I do not extend. And if there are no major pieces and more then 3 ranks separate the pieces, I do not extend.

This reduces the game tree, but the database hits are still very high.

So, this may allow Vortex to have "amost" a 7-piece database with this one litte trick. Time will tell if this will work.