I’ve been listening to some podcasts by Johns Gruber and Siracusa. Good stuff, well reasoned. One show talked about the first computer one of them had. These are both Mac enthusiasts, so it is no surprise that these were Apple machines. It got me thinking about my first machine, my first home machine.
My first machine ever was MIT’s IBM 7094. It used the CTSS time-share system and was an exact replica of a machine at NSA, ‘no such agency.’ But what we programmed on more experimentally in those days were PDP-11’s with punched paper spools. Boring stuff actually, when it comes to the hardware. Yes, I know everyone was celebrating the Ken Olson revolution that brought the minicomputer revolution. But the difference for us was slight: we had access to all the computing we wanted. All the companies involved and most of the key players are now dead. No story.
The computers that mattered were the ones we made with solder flux entering our bloodstream, the ones where it mattered if the resistors cannot not be matched and the transistor betas calibrated... the ones we could literally play with all the way from the last turtle. That was an era...
My first home computer was a Heathkit H-8 that I bought in 1978 as my family moved into our new house.
The Clunky Beast
This was a pile of components you bought from one source and soldered together. The Heath company had gotten into the kit business by selling ham radio kits, among the best sets you could get at any price. My roommate at MIT, son of a German soldier, had one in our dorm room. You had to ‘send away’ for it.
The machine was designed around the then new 8080 chip and programmed over a custom-written operating system, HDOS. Some programming and monitoring could be accomplished by a panel on the front that had a keypad and an LED (rather than the then common Nixie tube) display. It had a 50-pin bus that held (I think) 8 cards.
I built a desk for it (which I still have), when it grew beyond the one cage to several linked by multiple ribbon cables and single conductors. Once the first cage was working as advertised, we set it up with a port of WordStar to HDOS, plus some games for the son.
The hardware ultimately ended up being four H8 cages with all slots filled. One contained interface and link cards, while three contained AMD 9511 math chips, four per card, six cards per cage: 72 chips. By that time, the master microprocessors were Z80’s and the cards were from a small outfit called Trionyx. They extended the machine tremendously, not only by upgrade and specialty cards, but by using gold-plated connectors. That gold was worth its weight in bits because the tin connectors of the original design made the machine frustratingly unreliable.
The goal by this time was to validate a specific theory by creating an ad-hoc pivoting Hilbert machine. It worked on the test case we gave it when modeling periodic vortices in an NP-hard problem space: a novel wind turbine. I kept that machine far beyond its useful life and eventually just tossed it, but not after I convinced someone to make some rackmounted cages to be paired with some Harris 24-bit machines that we were using for a spooky project project on Navy ships. The assembly language by that time had evolved into a Forth-based domain specific language, roughly emulating the Ada we were interfacing with on the Harris machines. It was created by a couple autistic kids I found in Portsmouth that we called Frick and Frack.
This was all independent of what would come to be called the personal computer and Charlie Chaplin’s huckstering.
Chaplin Used to Market IBM PCs
And the work segued into multi-threaded, ‘folded’ semantics on something called SGML that predated the web. By the time we retired that generation, the system was called ALice: Ada-Lattice integrated conceptual environment.
Much of the work in other areas of this site follows from that earlier effort, and when I dream of the work, sometimes I can see the flux swirls and the crude printed circuit patterns of those Trionyx-based parallel tensor machine.