Collector's Corner..On Dedicated Chess Computers...

Post by **Steve B** » Sun Jul 29, 2007 10:12 am

Saitek Released a total of 9 different wooden boards over the years dividing them into three categories each with three models in them
the smaller boards consisted of the Capella , Astral and Virtuoso
the midsized wooden boards were the Stratos,Corona and President
and finally the large wooden boards were the Leonardo,Galileo and Renaissance

in the mid-sized category the Stratos and Corona look exactly the same but there are differences in some functions and keys and most importantly the Stratos is pressure sensory with 16 leds to indicate move output while the Corona is an auto-sensory with 81 leds(one on each square)
both computers are rated 1813(Selective Search) with programs by Kaplan
the President also an auto-sensory boasts an Engine by Morsch rated 1993(SS) and is a slightly larger board then the other two
all three boards had an opening book of 100k Moves

as the mid-sized boards were due for testing i matched each board against the Novag Citrine
much has been written about the Citrine on these pages and else where on the net so i will quickly mention it is rated 2084 at 40/2 and 2129 Game 30(.INFO)and has an opening book of 12k

First up was the Corona

as the Corona is rated several hundred points less then the Citrine i was expecting a massacre
however as will be seen in the game,the huge difference in the opening book size more then compensated for the rating difference

In the Game the Corona had White and reached 4 ply typically, pondering about 250 nodes per second while the Citrine hit 7-8 ply and 4000 nodes
the Citrine however was out of book first at move 3!

[[White "Saitek Corona"]
[Black "Novag Citrine"]
[TimeControl "Game/30Min."]

1.e4 d5 2.exd5 Nf6 3.c4 e6

[fen]rnbqkb1r/ppp2ppp/4pn2/3P4/2P5/8/PP1P1PPP/RNBQKBNR w KQkq - 0 4[/fen]
Citrine was expecting 3.d4 and was already on its own flashing a PV of -.54
down a pawn by the third move the stronger computer simply cannot recover

4.dxe6 Bxe6 5.d4 Bb4+ 6.Bd2 Bxd2+ 7.Qxd2 O-O
8.Nc3 Re8 9.O-O-O Bf5 10.Bd3 Bxd3 11.Qxd3 a6 12.Nf3 b5 13.cxb5 Ng4
14.Rd2 Nf6 15.bxa6 Rxa6 16.Re2 Rxe2 17.Qxe2 Re6 18.Qd3 Nc6 19.Qc4 Qd6 20.d5 Ne5
21.Nxe5 Rxe5 22.Rd1 Rf5 23.Nb5 Qxh2 24.Qxc7 Qxc7+ 25.Nxc7 Rxf2 26.Rd2 Rf1+ 27.Kc2 Re1
28.a4 Re4 29.Kb3 Nd7 30.Na6 f6 31.a5 Kf7 32.Kc3 Ke7 33.d6+ Kd8 34.b4 Kc8 35.Nc5 Re1 36.Nxd7 Kxd7 37.Kc4
Ra1

[fen]8/3k2pp/3P1p2/P7/1PK5/8/3R2P1/r7 w - - 0 38[/fen]

still down that same pawn but of course now with a totally lost game due to the Corona's passed Qside pawns
strangely the Citrine has no sense of just how bad its position is here and evaluates it at only -2.29 while the Corona correctly scores it +4.06

38.Kb5 Ra3 39.a6 Ra1 40.Kb6 Kc8 41.b5 h5 42.d7+ Kd8 43.Kb7 h4 44.a7
h3 45.gxh3 *

FINAL POSITION

[fen]3k4/PK1P2p1/5p2/1P6/8/7P/3R4/r7 w - - 0 45[/fen]
the Citrine resigned here

some what shocked at this rather unexpected turn of events i loaded up the less expensive Stratos and gave the White pieces to the Citrine..
surely this will be a piece of cake for the Citrine...

Dont Change That Dial Regards
Steve

mclane · Post by **mclane** » Sun Jul 29, 2007 10:36 am

steve, i am in the moment much interested in the NIGEL SHORT.

maybe we could/can produce more data about the Nigel short (i am testing it in the moment to find out where it is in the rating lists).

there were

Code: Select all


Mephisto MMV Turbo 16-18 MHz	                2125
Mephisto MegaIV 10 MHz		                   2070
Mephisto Milano			                       2016
Mephisto Academy		                         1999
Mephisto MMV+HG550		                       1988
Mephisto Monte Carlo IV LE	                  1987
Mephisto Mega IV		                         1980
Mephisto Polgar			                       1972
Mephisto Monte Carlo IV		                  1961
Mephisto MM IV+HG440		                     1943
Mephisto Rebell 5.0		                      1871

where in this list would the Nigel Short belong to ?

Post by **Steve B** » Sun Jul 29, 2007 10:55 am

mclane wrote:steve, i am in the moment much interested in the NIGEL SHORT.

maybe we could/can produce more data about the Nigel short (i am testing it in the moment to find out where it is in the rating lists).

there were
Code: Select all
Mephisto MMV Turbo 16-18 MHz	                2125
Mephisto MegaIV 10 MHz		2070
Mephisto Milano			2016
Mephisto Academy		                1999
Mephisto MMV+HG550		                1988
Mephisto Monte Carlo IV LE	                1987
Mephisto Mega IV		                1980
Mephisto Polgar			1972
Mephisto Monte Carlo IV		1961
Mephisto MM IV+HG440		1943
Mephisto Rebell 5.0		                1871
 
where in this list would the Nigel Short belong to ?

hi Thor

yes a very special computer
also called the Super Milano
it should be about 30 elo higher then the normal Milano
SS shows it rated at 1987

so in your list it should be about 2045

time for a link regards:

http://www.flickr.com/photos/10261668@N ... 923814103/

Steveb

mclane · Post by **mclane** » Sun Jul 29, 2007 10:58 am

i have a 10 mhz nigel short here. but my problem is, that i do not have the 5 mhz version to relate it with the normal milano or the polgar or whatever.
i will see what i do for it.
thanks steve.

thanks for the nice image

do they say anything in the manual what is different to the milano,
or is the manual just the milano one with a different header ?!

has anyone done the Bednorz-Tönissen with the NigelShort too ? so that we could relate the results with each other ?!

Post by **Steve B** » Sun Jul 29, 2007 12:55 pm

the manuals are the same and all of the features are identical

Ed Schroeder discussed the Nigel Short on the CCC about a year ago
at first he did not even remember that he was the programmer for it

but after awhile it was mentioned that the only difference between the computers was the program itself..he made some program revisions which produced a slightly higher elo(30)

Best
Steve

mclane · Post by **mclane** » Sun Jul 29, 2007 2:01 pm

yes i remember these statements.
i have found the BT2450 of Milano : 1956 BT-ELO.

will make some positions with the nigel short...

i wonder if the mephisto nigel short is the strongest 8-bit chess computer.

the BT results so far were strange.
different from polgar, mm5 and milano.
strange.

Post by **Mark Uniacke** » Sun Jul 29, 2007 6:12 pm

Who programmed the Corona? was it the same guy who did the turbostar 432 and Simultano?

His name is on the tip of my tongue but I can't quite remember it!

I have both of those and they play some interesting chess. I believe they use a selective search but probably Thorsten will remember more about that.

Also what was the cpu & speed of the Corona?

Mark

Post by **Steve B** » Sun Jul 29, 2007 6:23 pm

Mark Uniacke wrote:Who programmed the Corona? was it the same guy who did the turbostar 432 and Simultano?

His name is on the tip of my tongue but I can't quite remember it!

I have both of those and they play some interesting chess. I believe they use a selective search but probably Thorsten will remember more about that.

Also what was the cpu & speed of the Corona?

Mark

hi Mark
as i briefly mentioned at the top of my initial post the Corona and Stratos were both programmed by Julio Kaplan

the Corona sported a 6502 running at 6 Mhz

Good Luck With Your New Forum Regards
Steve

mclane · Post by **mclane** » Sun Jul 29, 2007 6:41 pm

julio kaplan, yes. wasn't he a very strong and young IM /champion
in his past ?

Post by **Steve B** » Sun Jul 29, 2007 6:52 pm

mclane wrote:julio kaplan, yes. wasn't he a very strong and young IM /champion
in his past ?

oh Yes

a fine player himself
like Rajlich,Uniacke,Levy and others

i think it helps to have a strong player as a programmer but it is not a necessity of course

Best
Steve

Post by **Mark Uniacke** » Sun Jul 29, 2007 8:11 pm

Thanks guys, yes Kaplan was a strong player and I think he tried to do some interesting things in his programming.

The program I still like is Mephisto 3 by Nitsche and Henne?

Thorsten had a very interesting article about how it picked its moves.

It searched very slowly at about 2-3 nps on my exclusive version which I think was a 1608 at 6Mhz ot something like that. It was not so good at blitz because it needed some time to get a level of tactical ability but it was very interesting.

I believe they made a faster and much stronger Exclusive S version running on 68000/8Mhz but I think that was the end for them as the Lang programs took over.

mclane · Post by **mclane** » Sun Jul 29, 2007 8:19 pm

kaplan had kind of planning algorithm kind of hash tables or killer moves or whatever. fact is that it did not always played the same moves.
sometimes when it had a computation situation before where it found a PLAN, it was able to make LATER a move.
if you than e.g. took the later position, set it up it was NOT capable finding this move. it needed a situation a few moves before and the computation process than.

IMO this was kind of hash in 8 bit, or he stored the MAIN-LINES from before, so he was able to CONTINUE the idea in later situations.

here the game:
1.Turnierpartie (40i120) Mephisto Milano: D++ Experimental 8Mhz

1.c4 e6 2.d4 d5 3.Sc3 Sf6 4.Lg5 Le7 5.cxd5 exd5 6.e3 c6 7.Ld3 0-0 8.Sge2 Sbd7 9.Dc2 Te8 10.0-0-0 Oh, unterschiedliche Rochaden. Sf8 Oh, herrlich. 11.h3 a5 12.g4 Sg6 13.f3 h6! Schön. 14.Lxg6 hxg5! 15.Ld3 Lb4 16.e4 Le6 17.a3 Le7 18.The1 b5 19.e5 Sd7 20.f4 b4 21.f5 Ach, wie dafür bezahlt. Entertainment! bxc3 22.Sxc3 Sxe5!! 23.dxe5 Lc8 24.e6 fxe6! Zum Teufel mit Zügen wie f6. Hier wird offensiv gespielt. 25.fxe6 Ld6 26.Lh7+ Kh8 27.Dg6 Wer hat schon Angst? Lxe6 28.Txe6 Txe6 29.Dxe6 Kxh7 Uff! 30.Kc2

Man / Frau meint alles sei vorbei. Df8 31.Se4 Lf4 32.Sc3 Dc5 33.Kb1 Tf8 34.Sa4 Dc4 35.Sc3 Db3 36.Ka1 Tf6 37.De2 Ld6!! Hier ist ein Plan am Reifen.

38.Dd3+ Kh8 39.De3

Keines Ihrer Geräte findet den folgenden Zug dieser Stellung, wetten ?
Wer findet das in welcher Zeit??

40.bxa3 Dxa3+ 41.Kb1 Db3+ Remis ? 42.Kc1 Nein, mehr !!! a4 43.Kd2 Aber wieviel mehr? a3!! Ganz viel mehr! 44.Tb1 Dxb1!!
NA? Überzeugt daß Kaplan Schach spielen kann?

45.Sxb1 a2 46.Sc3 a1D 47.Kc2 Df1 48.Sd1 Dc4+ 49.Kd2 Df4
50.Kd3 Dxe3+ Immer muß dann getauscht werden.

51.Kxe3 c5 52.Ke2 d4 53.Kd3 Tf3+ 54.Ke4 und der Rest war Schweigen! 0:1

without the computation on Bd6 in move 37. it was NOT capable to find
/play 40.bxa3.

only if it found Bd6 it later played 40.bxa3.

And all this in 8 bit chess computer !

nitsche + henne. yes. very interesting commercial program.
i am still in contact with thomas nitsche. he again works with elmar henne in munich in a company (only they do not program chess anymore).

there has never been again such an approach like mephisto III.
its a pity we "lost" them on the way.

not only that Lang took over. there were SEVERAL things.
e.g. people believed that THERE way of doing computerchess was WRONG and people like ossi weiner, friedel and steinwender critisized them,
and called Thomas "mad genius" and stuff like that.
although his ideas were concrete. SO concrete that you could win championship with it and even make a commercial dedicated unit
with very small hardware.

IMO nitsche was TOO EARLY.
later, with the bigger hardware, they could have researched much better
in their field. imagine the thing that Mephisto III could print out the whole search tree ! with all evaluations.

e.g. friedel/steinwender believed that nitsche cheated in bratko-kopec.

here mephisto III search tree for Nf6:

Post by **Steve B** » Sun Jul 29, 2007 9:45 pm

Mark Uniacke wrote:
I believe they made a faster and much stronger Exclusive S version running on 68000/8Mhz but I think that was the end for them as the Lang programs took over.

Yes they did
but it was called the Mephisto Excalibur and came in its own special board and it was also the very first dedicated chess computer ever released for sale to use the 68000 processor

very rare computer indeed!:

http://www.flickr.com/photos/10261668@N ... 923814103/

Stuff Legends Are Made of Regards
Steve

Post by **Mark Uniacke** » Sun Jul 29, 2007 10:42 pm

Thanks Thorsten, I knew you would have the info.

So did they cheat on the BK tests?

I really love the idea of a chess program where you can see the whole chess tree and understand it. Incredible especially considering the era these guys were working in.

It reminds me of the Paradise knowledge based lisp chess program. That did very well for the day on the first 100 WAC positions. I wonder if anyone out there has the original ACM article on Paradise?

Steve, that was a very rare chess computer and ahead of its time in many ways. I believe it came equal first in Glasgow in 1982 or 83?

Post by **Steve B** » Sun Jul 29, 2007 11:00 pm

Mark Uniacke wrote: Steve, that was a very rare chess computer and ahead of its time in many ways. I believe it came equal first in Glasgow in 1982 or 83?

Yes!
Equal first in 84 with Fidelity, Concess, and an early version of Langs Psion
Mephisto released the weltmiseter program soon after in its own module shown here:

http://www.flickr.com/photos/10261668@N ... 923814103/

Ultra Rare Regards
Steve