Slashdot Mirror
DIY CPU Demo'd Running Minix
DeviceGuru writes "Bill Buzbee offered the first public demonstration of the open-source Minix OS — a cousin of Linux — running on his homebrew minicomputer, the Magic-1, at the Vintage Computer Festival in Mountain View, Calif. The Magic-1 minicomputer is built with 74-series TTL ICs using wire-wrap construction, and implements a homebrew, 8086-like ISA. Rather than using a commercial microprocessor, Buzbee created his own microcoded CPU that runs at 4.09 MHz, and is in the same ballpark as an old 8086 in performance and capabilities. The CPU has a 22-bit physical address bus and an 8-bit data bus."
Linus copied Minix. Well known fact !!
I wirewrapped a computer together back when building your own hardware was about the only option, and it wasn't a fun experience. I can't imagine actually wanting to do it, but to each his own.
Linux was originally host compiled on Minix. It's original filesystem was Minix compatible. Linus originally announced Linux on the Minix newsgroups. They're both *nixen. I think that cousin is a pretty good description. Though maybe Linux as a bastard child would be more accurate.
I'm not going to link to it, because I don't want this hobby project to go up in flames, but if you follow the links to the website of the guy who built it, you would find that he's actually running a webserver on it.
Even more importantly, can't you guys realise that none of these jokes are funny?
The reason the Magic-1 isn't in service as a webserver is that, at the moment, Bill's showing it off at the Vintage Computer Festival.
Also, many early linux boot floppies contained a minix filesystem. I recently had to put a Slackware box online specifically so I could read some old minix filesystem floppies I made back in the mid 90's.
Minix back then was open source (non-TM version) but you had to buy the textbook to legally use a copy. Now it's open source and the latest version is quite respectable.
It has an open guest telnet and a webserver running (when he's not working).
It is, apparently, slashdotted right now just through people following the chain of links and finding it.
How much slashdotting does it take to take a 4 MHz machine to it's knees?
And more importantly, did it stay UP during the slashdotting, but just get as slow as Unreal on a 100 MHZ machine w/o 3D-hardware acceleration?
You are right, but that doesn't help when you're such a rude asshole.
You must be new here.
In small quantities, they're more expensive than wire-wrap, although it depends on what your time is worth. Of course you can spend your time laying out a PCB or spend it doing wire wrap. I'd do the PCB. Especially if I needed more than one. Is wire-wrap better for multi-layered circuits, or something? No, PCBs are superior. Of course there are little details that are quite important, and if you don't know what you're doing, you can easily design a PCB that doesn't work.
I think the guy did it with wire wrap because it's retro. Hey, whatever floats yer boat.
From my understanding, MIPS would be a lot more complex on the processor side of things than basic 8086. Remember we're not talking about Pentium-4 x86 or AMD64 here, we're not even talking about the venerable old 386. we're talking about the real 8086 which was pretty basic. I don't think it had any pipelined instructions, which is something you'd have to deal with in MIPS.
Random and weird software I've written.
If he were a typical Slashdotter, I'd agree ... but he apparently has a wife and a couple of kids.
The higher the technology, the sharper that two-edged sword.
I don't think that you have to make a pipelined CPU, just because it implements a MIPS-like ISA..
I know that a few instructions (the branching slot) only works on a pipelined implementation, but it isn't necessary to make a fully compatible MIPS.
And even a 'basic x86', is quite complicated with its instruction with a varying length..
and minix copied unix, which copied multix.
Windows copied Macintosh, which copied the Lisa (also from apple), which copied the Xerox Alto and Star, which copied the oNLine System (1965).
If by "copied" you mean "got ideas from." In science this is not considered cheating. It is considered doing your homework. If you don't look at other successful designs before making your own, there can be no progress. We'd end up reinventing the wheel 100 different broken ways, instead of coming up with better and better iterations on the same theme.
Linux was "inspired" by Minix, but succeeded in its place because of higher performance and a more open development environment.
Babbage's engine was hard to make *then* because they didn't have good machining capabilities. Nowadays you can make a small difference engine out of LEGO blocks -- it doesn't get much more POTS than that. (notice the geek cred in the domain name, btw.)
It's surprisingly easy to draw wire. I've done lots of it. The original stuff was done without drawplates: they filed a notch in a plate, then put a second plate against it, clamped them firmly, and pulled, then used the next, smaller notch. You get a half-circle of wire that tends to curl but it's doable. Insulation is *much* harder -- making something that's flexible, tough, and has a reasonable dielectric, and getting it to stick to the wire, is *hard*. Drawing 30 gauge copper tubing is easy in comparison.
Voltaic piles are easier to make than Leyden jars. If you're bored, you can light an LED with a stack of small pieces of aluminum and pennies, separated by lemon-juice-soaked paper towels. It took me about 7 of each to get a red LED to light. If people had known what to do they could've made voltaic pile batteries in Egyptian times -- separate copper and silver chunks with spit- or saltwater-soaked papyrus sheets.
There were early relays made from glass tubes with wires and piles of steel filings. An electric charge on one wire attracted filings, which bridged to the other wire. You could use those to make primitive high-power diodes as well, by messing with the geometry of the wires -- again, stuff that any culture with some competence in glass could've done (and that's pretty old.) The problem was always one of basic research and not knowing what to try.
Nostalgia's not what it used to be.
At least not if you want tubes that might operate long enough for the computer to actually get through boot sequence.
While tubes are simple in concept, the amount of chemistry, metallurgy, and material science that went into making reliable vacuum tubes was simply astounding. Particularly for applications involving hundreds or thousands of tubes (like computers), achieving very high tube reliability is key to getting the computer to run long enough to actually crank outa few calculations before a tube fails.
Tubes that were designed for computer service needed ultrahigh purity metals, particularly nickel for the cathodes. The level of vacuum needed is FAR higher than you could get with a simple mercury siphon pump (think turbomolecular or oil diffusion pump). Exotic metallurgy and coatings are needed to produce grids and plates that don't emit their own secondary electrons. Cathode coating chemistry was jealously guarded by most manufacturers, and also critical to decent life.
All of this stuff is pretty much a "lost art" these days, and it is likely that nobody will EVER be able to duplicate the quality of the best tubes of the past, as most of the people who did it are now dead. While you can make a triode that will function as an amplifier with rudimentary glassblowing skills, making a tube that will reliably work in a high speed pulse switching environment such as a digital computer takes a great deal more knowledge and infrastructure.
Tube manufacturing was every bit as complicated as semiconductor manufacturing is today.
Remember "News for Nerds, Stuff that Matters"? Help make it a reality again! http://soylentnews.org