Chris Whittington's Chess System Tal!

[mclane]

no thank you tord. i am too stupid for changing the game.
i will stay in the chess area. and not try another complication

[chrisw]

Hi Tord,

Yes, I produced a very agressive Shogi engine in about three months. I think it was built on the back of the routines in CSTal3 (not the bitboard ones!) in 1998/99 or so.

What has happened was that OS had licenced Jeff Rollason's Shogi to its Japanese distributor, the revenues were large and important (PC and Play Station). There was strong Shogi competition and Jeff was struggling a bit at one stage. OS had very little control over what he did since nobody understood, and there was always the threat of losing the Japanese deal to competitors. So I attacked the Shogi problem for two reasons, one to put a rocket up Jeff's backside and the other to try and understand the basis of Shogi and Shogi engines in order to manage the project better.

The result was immensley agressive, the underlying routines of movgen, move, unmove are fairly simple to change, so really all the work for a chess programmer is evaluation. I just went for massive king attack and used standard piece values for the different Shogi pieces.

The Japanese distributor came over and watched it play a couple fo games and liked it. Jeff had got everything back together again by then, so I didn't do any more with mine and the source probably sits around somewhere.

That's a shame. I guess somebody else owns this code too, and that there is little or no hope of ever seeing it released in any form?

Any, the moral is that Shogi is not very difficult for a chess programmer who already has a working chess program.

If it were only the engine, it wouldn't be too bad. The problem is that I have to write a GUI as well. The GUI for my chess program is messy and poorly written, and I think modifying it to handle shogi would be about as much work as writing a completely new GUI from scratch. I would also have to do some work to port the low-level bitboard stuff to shogi, which isn't really technically very hard, but awfully boring.

Tord

Ah, yes, well, I had a great GUI programmer, Matt Grogan, working with me as general support programmer. Matt joined direct from University College Oxford and was a great guy, really good and fast at getting stuff done. So, I did engine and Matt, who had already done the entire GUI for the Jeff Shogi (Jeff was also an engine guy, his test bed was operating out of a DOS box), was able to cannibalise that GUI and wrap it round my engine - full graphics, everything.

I guess that source is somewhere.

Best wishes,

Chris

[swami]

What's the result so far in match Romi vs CSTal 3, Thorsten?!

Can you also post games?

[Tord Romstad]

no thank you tord. i am too stupid for changing the game.
i will stay in the chess area. and not try another complication

OK, perhaps not a bad idea. If you try to do everything at once, you end up being pretty bad at everything you do, just like me.

I'll extend my offer to everybody else who want to learn shogi: I'll play against you with a rook handicap. The easiest way to do it is probably to register at http://www.brainking.com (it's free), where you can read the rules and play correspondence games. Click on the "Player List" in the frame on the left, choose the player named "Tord", and challenge me to a game of Japanese Chess (preferably an untimed game).



Tord

[mclane]

first game win for romi:

[Event "core2quad 6600"]
[Site "40/120"]
[Date "????.??.??"]
[Round "1"]
[White "CST2.03"]
[Black "RomiChessX64P3k"]
[Result "0-1"]

1. d4 Nf6 2. Bf4 d5 3. e3 e6 4. c3 Be7 5. Bd3 Nbd7 6. Nd2 c5 7. Ne2 Nh5 8. Be5
c4 9. Bc2 O-O 10. Qb1 g6 11. g4 Nhf6 12. Bf4 Nxg4 13. Rg1 f5 14. h3 Ngf6 15. b3
b5 16. Bh6 Re8 17. a4 b4 18. bxc4 bxc3 19. Nxc3 Qa5 20. Nb5 Bb4 21. Qd1 Nb6 22.
Rb1 Nxc4 23. Rxb4 Qxb4 24. Nc7 Bd7 25. Nxa8 0-1

permanent brain was ON. Romi played with mainbook in arena.
CSTAL used no book.

Now the same opening but this time against "bitboard CSTAL3".


Bitboard CSTAL3 is doing 300.000 NPS (or 3 times faster NPS than CSTAL2.03), but Romi of course still 11 times faster.

[chrisw]

first game win for romi:

[Event "core2quad 6600"]
[Site "40/120"]
[Date "????.??.??"]
[Round "1"]
[White "CST2.03"]
[Black "RomiChessX64P3k"]
[Result "0-1"]

1. d4 Nf6 2. Bf4 d5 3. e3 e6 4. c3 Be7 5. Bd3 Nbd7 6. Nd2 c5 7. Ne2 Nh5 8. Be5
c4 9. Bc2 O-O 10. Qb1 g6 11. g4 Nhf6 12. Bf4 Nxg4 13. Rg1 f5 14. h3 Ngf6 15. b3
b5 16. Bh6 Re8 17. a4 b4 18. bxc4 bxc3 19. Nxc3 Qa5 20. Nb5 Bb4 21. Qd1 Nb6 22.
Rb1 Nxc4 23. Rxb4 Qxb4 24. Nc7 Bd7 25. Nxa8 0-1

permanent brain was ON. Romi played with mainbook in arena.
CSTAL used no book.

Now the same opening but this time against "bitboard CSTAL3".


Bitboard CSTAL3 is doing 300.000 NPS (or 3 times faster NPS than CSTAL2.03), but Romi of course still 11 times faster.

Does CSTal3 have any meaningful evaluation function?

[mclane]

is this a philosophical question ?
as far as i remember you made the new engine in different kind of stages.
those stages include step by step different stuff from OLD CSTAL2.03 ported into the new engine.

those different stages were named primitive engine 0,1,2,3,4,5,6,7.
as far as i can see it the version 0 used the most ported knowledge/stuff from CSTAL2.03 while the smallest engine (7) used almost nothing.

7 makes most NPS
while 0 makes smallest amount of NPS. Still 0 is 3 or 3.5 times faster than CSTAL2.03.

IMO the bitboard version still has a different search (whatever evaluation function beeing used) .

from time to time in those days you sent me new LATEST engines.
IMO the engine i refer to as bitboard CSTAL3 was from 07.09.1999.

there is a file in the source code with different switches for the different engines. in this file the engines gets more and more ported stuff from CSTAL2.03 if the flag is set to 1.
from what i see i would say that the amount of useful knowledge that can be set to 1 is ok. i did not look into the code to see if the names of the functions or procedures are WORKING or only dummys.


btw. bitboard CSTAL3 plays different, game now develops:

[Event "core2quad 6600"]
[Site "40/120"]
[Date "2008.05.24"]
[Round "?"]
[White "Bitboard CSTAL3"]
[Black "RomiChessX64P3k"]
[Result "*"]

1. d4 Nf6 2. Bf4 d5 3. e3 e6 4. Nd2 Be7 5. Bd3 O-O 6. c3 c5
7. dxc5 Nc6 *

[chrisw]

is this a philosophical question ?
as far as i remember you made the new engine in different kind of stages.
those stages include step by step different stuff from OLD CSTAL2.03 ported into the new engine.

those different stages were named primitive engine 0,1,2,3,4,5,6,7.
as far as i can see it the version 0 used the most ported knowledge/stuff from CSTAL2.03 while the smallest engine (7) used almost nothing.

7 makes most NPS
while 0 makes smallest amount of NPS. Still 0 is 3 or 3.5 times faster than CSTAL2.03.

IMO the bitboard version still has a different search (whatever evaluation function beeing used) .

from time to time in those days you sent me new LATEST engines.
IMO the engine i refer to as bitboard CSTAL3 was from 07.09.1999.

there is a file in the source code with different switches for the different engines. in this file the engines gets more and more ported stuff from CSTAL2.03 if the flag is set to 1.
from what i see i would say that the amount of useful knowledge that can be set to 1 is ok. i did not look into the code to see if the names of the functions or procedures are WORKING or only dummys.


btw. bitboard CSTAL3 plays different, game now develops:

[Event "core2quad 6600"]
[Site "40/120"]
[Date "2008.05.24"]
[Round "?"]
[White "Bitboard CSTAL3"]
[Black "RomiChessX64P3k"]
[Result "*"]

1. d4 Nf6 2. Bf4 d5 3. e3 e6 4. Nd2 Be7 5. Bd3 O-O 6. c3 c5
7. dxc5 Nc6 *

The search was brought more up to date I think. It relied more on the quiescence function and didn't try doing any static material evaluation (I think).

[Michael Sherwin]

imo chris lost interest in writing another chess program.
something that has happened to many others in the scene.

Romi is doing between 3.300-3900 KNPS while CSTAL is doing 100 KNPS
on the Core2quad.

That surprises me that my single threaded program runs that fast! Your machine must be overclocked.

[Michael Sherwin]

Hey Chris,

I am 50 and I still have one or two more engines left in me. I just need some time to work on them.

About your old stock market data. I thought about looking for some core groupings that (almost) always move together. Then finding single stocks that move in a certain direction a high percentage of the time some days, weeks or even months later. Ofcourse this is simple thinking and probably has been tried many times.

Anyway I am starting to detect a bit of an itch in you for chess programming that may not be completely scratched yet!