Just a flip flop and an OR gate.

> until you add a whole bunch of external hardware

Why mind this if the hardware to add is the same level of technology like Z80 itself? I would be glad to build a computer as advanced as possible combining some Z80s and additional HW.

> Why mind this if the hardware to add is the same level of technology like Z80 itself?

But it's not "the Z80's protected mode" - it's "the Z80 + an extra circuit to emulate protected mode". It's a super clever hack, yes! But filthy clickbait also.

There’s a small irony in your clickbait complaint containing a misrepresentation of the content.

It's not uncommon for chips to support configuring different features by wiring in external components. Though usually they are passive components, this isn't completely absurd.

I mean, true, but "a whole bunch" is kind of misleading. It's a really clever hack using really a tiny amount of extra gates. It's still a very cool and innovative hack that nobody really had thought about before.

Reminds me of the printer interrupt hack Darklite/Offence used for their Tiki-100 demos.

The Tiki-100 is an obscure Norwegian 8-bit computer from 1984, also Z80-based. It had nice graphics for its time, with 16 colours from a 256-colour palette. However, it was very hard to use productively, since the machine had no video sync interrupts.

What the guys in the demogroups figured out, was that it did have parallel port interrupts, and with a tiny hardware mod (just a wire from the expansion port to the parallel port), they could "steal" those for Commodore 64-style raster interrupts.

Heck, one could connect a Commodore 64 to the Tiki and get synced frames.

It certainly has been thought of before. Using external circuitry to remap the address space was common with lots of different processors, and once you have that (and a way to jump to a controlled address from hardware) it's a logical next step.

That it wasn't something usual with microcomputers of the time wasn't because it would have required some uniquely clever design. Rather it had to do with them being single-user/single-task, with simple operating systems where programs were expected to have full access to the machine.

Yes, there were lots of 8-bitters that used bank switching. But I think you have the causality somewhat backwards. 8-bitters didn't lack memory protection because they didn't have operating systems that could use it; rather, their operating systems couldn't use it because it wasn't there. Multitasking was of course useful, then as now, and sometimes people did do multitasking on them despite the generally limited memory; it was just less reliable because programs could crash each other. Multitasking was common in real-time control, which is how THERAC-25 happened (on a non-8-bit machine, but due to a race condition between threads).

A common use for multitasking in the 8-bitter era was print spooling; it was very inconvenient to have to wait for your document to finish printing before you could continue editing! People frequently bought external hardware printer spooler boxes to connect between their computer and printer because of this. Similarly, duplicating floppies was pretty inconvenient because you couldn't do anything else while it was happening, even if you had four floppy drives like I did, with literal megabytes of storage space available to read porn or play video games while the floppies were copying.

Yeah, but as the author says, it renders the whole maskable interrupt system useless… so it is not something you would do since it would undermine too many useful parts of the chip.

Using external signals for some sort of protection isn't innovative. This is a documented feature of the Intel 8086: the S3 and S4 pins expose which memory segment is being accessed: code, data, stack or extra segment.

Sure, but if memory serves me correctly, many early workstations used a microprocessor with an external MMU. It is misleading in the sense that it is not part of the Z80, but it is in the character of the times. (And MMUs are not the only CPU feature that was originally external. The FPU is an example that more people will be familiar with since it was an upgrade path available on many personal computers.)

I don't think MMUs were common on machines that ran on a Z80 however.

The home-brew pager (MMU) of the MIT-MC PDP-10 KL10 was appropriately called "KL-UDGE". (KL was the prefix of the PDP-10 model).

https://en.wikipedia.org/wiki/Kludge

It's on display at the Paul G. Allen Living Computer Museum.

https://sites.google.com/site/mthompsonorg/Home/pdp-10/pdp-1...

https://donhopkins.com/home/catalog/images/mc-console.jpg

Lars Brinkhoff has taken photos of KL-UDGE blinken lights panel and naked wire wrap boards:

https://twitter.com/larsbrinkhoff/status/1128601896619524097

>Photo 6/7: KL-UDGE lights panel

>Photo 7/7: KL-UDGE boards. Real-time time of day clock?

Was on display. The LCM closed due to COVID and there's no indication that it's coming back, sadly.

The Commodore 128 had an 8502 CPU, a Z-80 CPU and an 8722 MMU.

Was a pretty popular machine - sold 5.7 units worldwide. But probably only a tiny fraction actually used the CP/M mode that used the Z80.

I don't know what's more concerning. The fact they only sold 5.7 units, or the fact someone managed to only buy .7 of a Commodore 128.

The sixth unit had CP/M disabled, so accountants wrote it down as 0.7 units.

Wow, I had no idea there was basically a second computer hiding inside of the C128. No wonder it cost so much.

Did you mean 5.7M?

For example a Programmable MMU was an optional add on for the Macintosh II (which wasn’t a z80 but serves to illustrate the time when they began to become available to home computers)