Ben Eater on Running a Breadboard at 1MHz

It works fine! This is a really useful video showing lots of basics of digital electronics. See https://eater.net/6502 for kits and videos.


Here's an idea for a little project based on Ben's kits. Make a high-speed bidirectional communications link between two asynchronously-clocked 6502 CPUs using a couple of blocks of, say, 2KB SRAM (on separate chips) which have their address logic controlled by the outputs of the 6522 PIAs in such a way that each CPU can write data onto one of the chips and then switch that chip onto the address and data bus of the other CPU. You will need a few tri-state buffer elements to do this.

How about doing a PCB link module with, say, an ICE-40 FPGA implementing this functionality and providing a pair of header-pins that could be used to bridge two breadboards?


The next idea is to design a PCB with four 6502 CPUs, each connected via one of these link modules to two others in a square, and having four sets of headers around the edge which could connect via the above link modules. This would be the moral equivalent of an Inmos Transputer.

Then add four more links so that these modules could be used as vertices of a hyper-cube.

See MicroCore Labs MCL65
The MC65 is an ultra-small footprint, microsequencer-based, 100% instruction-set compatible, cycle-exact NMOS 6502 core that can be implemented in any FPGA or ASIC technology which can utilize as little as 252 LUTs (0.77%) of a Xilinx Spartan-7 FPGA. It has also been ported to a Xilinx Spartan-3 device where it uses about 10% of the part.
... demonstrated here:


... and for some 6502 backround, see:


... and:


On 6502 design documents that turned up in '95, at 23 minutes 26 seconds, see Bill Mensch's patents, including the one for the 65C02: US4652992A.

At 29 minutes 57 seconds the process of emulating the layout at the NMOS level might be one which could be amenable to a stochastic search for solutions to differential equations. See The Right Way to Formalise Mathematics. Incredibly, this chip can be emulated at the semiconductor junction level in JavaScript! See from 31 minutes 33 seconds.

Now for the next phase, look at how Digital made the DEC PDP-11 CPU:


See in particular this Unibus driver:


Here is the TI datasheet:


See how closely packed the boards were in these unibus systems:


How much heat is generated by the motors and power supplies needed to maintain the high airflow through these systems? This depends on the ambient air temperature, of course. What does the ambient air temperature in a machine room depend upon? What does this cyclic dependency tell you about the critical feature of energy use in large-scale computing systems? What does that tell you and HRH Charles, Prince of Wales, about the airline industry and the UK economy in general? See Why Britain is Washed Up. See also this:


See Element 14 Learning Videos for some ideas for longer range interconnects. See Destin on Re-engineering Supply Chains for an idea of how to get this into production. Oh man, the good old days are back again! 😀💓


Here is a video showing more of the system integration, including the front-panel loader, bus terminators and peripheral device boot PROMs.


My idea, around this time, was that anybody could manufacture these 'main-frames' with basic power supplies and cooling options, and a bus backplane, and a blank space for front-panel hardware, and there could be any number of specifications for the signalling characteristics on the bus, so different manufacturers could build different cards for peripherals, memory, processors etc, and computers would be something you could put together from parts supplied by any company that was capable. But that never happened, I suspect because there was just too much profit to be made selling whole integrated systems. But something went wrong with that idea too, ...


If you don't believe this is real, .... well, ... what is real? Betty and Gordon Moore Library at Cambridge Centre for Mathematical Engineering Sciences? See Lori Hoping for A Big Green Comet.


Edsger W. Dijkstra on learning computer science from your mother while in the bathtub, ...


See David Wheeler's algorithm for computing logarithms at 12 minutes 17 seconds (737 seconds) and Thanks for BZIP2. See also Burrows–Wheeler transform and think about some of the advantages which computing this using the binary representation of strings might confer. See Cambridge Computer Laboratory and Toby on Logarithms in Rob Boss style:


At 20 minutes 33 seconds, what does this have to do with what he learned about the geometric mean and the definition of a number (being something other than one, i.e. the base of a logarithm function) from his mother whilst sitting in the bathtub at age 8? See:


... and:


... and:


For some hints on how to test digital logic designs using arithmetic on finite fields, see How To Be A Genius Part II and I Seem To Have A Problem With Google.

At 28 minutes 43 seconds, on symmetries under invariant functions, see How To Be A Genius Part II and :


On wife swapping invariants, see this:


... and on symmetric representations of unordered pairs, see:



So now you know ...


... why window cleaning really is important: 'Er Maj. Puts George Gallow[s w]ay in Mind of Vera Lynn.

This leads to the advanced course, where we study the thermodynamics of this process and look for a new regime which could support cryogenically cooled NMOS chips operating at frequencies of 100s of GHz and maintaining 1k linearly independent states on one 1986 CPU die. See Element 14 Learning Videos. Think I'm joking? I am. Pity about the US National Helium Reserve going up in smoke Bill Clinton. That stuff could have saved a shit-load of coal used to fire power generators to run air-conditioning units in Microsoft, Amazon, Google datacenters, couldn't it? Here's Freeman Dyson on the value of being able to publish partially complete or inconclusive results of scientific work:


Now listen to Dijkstra at 55 minutes 23 seconds on Galois Connection, up to 55 minutes 43 seconds where he finishes: "Sorry that I cannot go into details." This seems to have been going on since the seventies. See Patrick Cousot & Radhia Cousot. Abstract interpretation: a unified lattice model for static analysis of programs by construction or approximation of fixpoints published in ACM-SIGPLAN Proceedings, in 1977. It started in the USA in 1968. See Hannah Arendt on Pathological Political Associations and An Electro-mechanical Display Element from 1969. And the problem is not just in computer science, it is in physics too, and presumably every other intellectual field worthy of the name. See Freeman Dyson Making Complete Sense of Modern PhysicsUS Embassy Tehran Siege 1979 and also George Galloway on Assange Extradition Hearing and Cambridge Computer Laboratory. On the question Dijkstra is asked at 55 minutes 56 seconds, because of this prohibition against teaching or using the idea of a Galois Connection, the issues of proving correctness of concurrent systems were solved in secret by Dijkstra, Milner, Girard, Reynolds and many others, and they were published as 'esoteric texts'. See I'm Still Looking for A Job.

See We Need Universal Access to Secure Global Telecommunications.

All this shit is all starting to look a bit STS52 to me!


Note the legend here and here. Now just wait to see what the cobwebs are made of and what they're for!

See What is Socialist Economic Development? and Hopi Prophecies and the Fifth World of the Australian Aboriginals.

Comments

Popular posts from this blog

Stefan Molyneux on Resource Based Economy

Nikola Tesla and the End of The World

Sabine Hossenfelder - Lost in Math