III-S Glasgow Prototype

Many of the chess computer programmers from the 80s had the opportunity to improve their programs as the hardware was evolving. Some examples are Richard Lang, Frans Morsch, Ed Schröder, Dan & Kate Spracklen, David Kittinger, etc. A few, like for example Richard Lang and Johan de Koning, can now even see their programs on modern chess computers, like the Millennium ChessGenius Exclusive and Millennium The King Element, respectively. On the other hand others, for one reason or another, did not have continuity. The case that has often interested me most is "Mephisto Project 3" (Das MEPHISTO 3-Projekt), developed by the Germans Elmar Henne and Thomas Nitsche. In the early 80's, among other projects, they developed together with Hegener & Glaser the programs for Mephisto I, II and III. The so-called "Brikett".

German article: Das Mephisto 3-Projekt

The last chess computer with this program was marketed in 1984 by Hegener & Glaser, for their modular system, under the name Mephisto 'S'. Coincidentally 1984 is the year when Richard Lang starts his collaboration with this company, winning the WMCCC celebrated in Glasgow with his Psion Chess program. Maybe this is why Elmar Henne and Thomas Nitsche ended their collaboration with the company?

The Mephisto 'S' module incorporates a 16-bit microprocessor, the Motorola 68000, running at 12 MHz, and runs the "III-S Glasgow" program. In different lists its ELO is slightly higher than 1700.

Motivated by a friend who is fond of chess computers, I decided to build a computer with a compatible hardware platform, to see what performance I would be able to get out of this program, just as programmers would have done at the end of the 80s, if they had continued to evolve their program with the rapid advances that were taking place at the hardware level.

We currently have this program emulated in MAME, and it can also be run for example in the Revelation module by Phoenix Chess Systems, which emulates the III-S Glasgow at 66 MHz. But my motivation was to build a real computer, without emulations, as it would have been done in the late 80's or early 90's, the golden age of chess computers.

I began by designing a prototype from the knowledge gained after repairing and reverse engineering multiple Mephisto modules with similar hardware using Motorola 68000, such as the Amsterdam, Roma, Dallas, etc. modules.

The first step was the design of the electronic schematics:

After reviewing and debugging it, I designed the printed circuit boards (PCB) on two levels, with a Process PCB at the bottom, and a Top PCB for the LCD, Keyboard and Buzzer. I then sent the designs to a PCB factory:

Once I received the PCBs, I proceeded to assemble the components and solve some small design errors: It's called a prototype for a reason!

My computer does not only include the engine module, but all other components necessary for autonomous operation (power supply, LCD, keyboard, buzzer, ...) as well as the necessary interface to connect directly to any sensory board made by Mephisto (ESB, München, Exclusive or Modular):

Once I had checked that everything was working perfectly at 12 MHz (same system clock frequency as in the Mephisto 'S' module) I started with the improvements. Because I used modern high performance ICs, fast RAM and ROM chips, etc. I was able to reduce the latency of the Data bus, and also to make other optimizations. Using Motorola microprocessors with DIP-40 package type, I managed to make it work stably at 20 MHz:

I then designed an adapter board to be able to use the fastest Motorola 68000's ever made (with PLCC package), and so I was able to increase the system clock frequency up to 25 MHz, running at full stability for days, without stopping.

As proof of the increase in performance over previous computers with this program, we can compare the number of positions calculated within 6 minutes, after playing 1.f3 at the analysis level (LE 9):

• Brikett Mephisto III = 630

• Brikett Mephisto III ‘Special’ 12 MHz = 1230

• Módulo Mephisto ‘S’  12 MHz = 4260

• Prototype III-S Glasgow 25 MHz = 8510

Finally, I made some small adjustments, such as replacing the LCD with a backlit one, and the prototype was ready to face other computers and check its level of gameplay:

     At Tournament level (40/2) it has beaten the Rebell 5.0 module, so I think it could be close to 1900 ELO.

The prototype had been a success!

This unit is currently installed in an ESB of my friend Wolfgang Zugrav.

I built another prototype for myself, with which to participate in tournaments.

And this is not the end of the story. We thought we would take advantage of this design to develop a more versatile computer...

Take a look at my project "GM Petrosian".