A more interesting platform to develop for?

[almaember]

Hello, I'm not sure if this is the right forum to ask, but I think this is the closest.
So, I realized that developing yet again "Unix clone for x86" was a really boring idea. That's basically what 90% of the people on this forum are doing.
I want to do something unique, for some old(-ish) computers, preferably 16-bit, or at most 32-bit, and something that I can either get or build actual hardware for.

I originally wanted to develop something for the ZX Spectrum (I know that the Z80 is 8-bit, but it supports 16-bit addressing and arithmetics so it might as well be 16-bit), but, as it turns out, that only has around 16K to 48K of RAM, which I guess is better then the ZX81's 1K but not enough to get too creative.

So I shortly thought about building a very simple CPU myself, which, although not completely impossible, would take a lot of time that I'd prefer to spend on programming, not building a computer.

I'd prefer to stay with the Z80 or a similar CPU if possible.
Do you know any more interesting platforms?

[BigBuda]

It's not old, but it is rather limited in resources... why not the RP2040? (the CPU in Raspberry Pico?)

[almaember]

It's not old, but it is rather limited in resources... why not the RP2040? (the CPU in Raspberry Pico?)

Doesn't look bad... although I tend to avoid anything ARM based for non-technical reasons (the way they use intellectual "property" annoys me a little). But if it's the best option I'll go with it.

[Octocontrabass]

Tandy Color Computer 3? It has a Motorola 6809, which is firmly 16-bit, but it also has 128kB of RAM standard.

It might be tough to get one at a good price though - I only have one because it was cheap at the time!

[nullplan]

Personally, I'm partial to the PowerPC architecture. Unfortunately, finding hardware is a tad hard. You can probably not get new hardware anymore. You can probably get an old PowerMac or PowerBook on eBay. Or, if you enjoy a challenge, look at listings for "Cleco mPro400SG" on there. You will get a system with touch screen, CF card slot, SD card slot, USB, the works. No sound, but that ought to be secondary (and can be added through USB). And the processor is an MPC5121. 32-bit, as stipulated. e300 core, based on the venerable PowerPC 603, but with two data breakpoints and two instruction breakpoints. And by default it ships with a version of OS-9.

[iansjack]

Tandy Color Computer 3? It has a Motorola 6809, which is firmly 16-bit, but it also has 128kB of RAM standard.

It might be tough to get one at a good price though - I only have one because it was cheap at the time!

+1

The 6809 is the most elegant of the early microprocessors. (Note that you would probably find it easier to find a Dragon 32 - at least in the UK - which was almost identical to the CoCo.)

Alternatively, any of the 68000 family would be fun to develop an OS for. Old Amigas, Ataris, or Macs are fairly easy to get hold of.

[rdos]

Other possible alternatives include Microchip PIC controllers. These have varying degrees of complexity. Another possibility is to use FPGAs. Many of these have processor cores in hardware. FPGAs, of course, are much more powerful than most processors, at least if you design time-critical parts with Verilog.

[linguofreak]

It's not "actual hardware" in the full sense, but the PiDP-11 is a scale-model of a PDP-11/70 front panel that encloses a Raspberry Pi running a PDP-11 emulator.

The x86 architecture itself has some little-used corners that are fairly interesting. You might try writing an OS for the 286 subset of the architecture (segmentation only, don't use 32-bit segments).

To be honest, though, the "actual hardware" requirement may be fairly limiting for you here if you're not wanting to do something "boring". A lot of older architectures were created while the industry was still figuring things out and before certain modern commonplaces had been standardized upon. They also had to make tough and creative choices about how to deliver the most value in an era when high transistor counts were expensive. Most hardware built recently enough to be readily available is built to get s&%t done in an era where transistor counts in the hundreds of millions to tens of billions are doable.

[almaember]

Okay, so thanks for all the answers everyone!

If someone was wondering, the primary reason that I'm trying to find something old isn't even that it's more interesting (although that's very important), but that it doesn't suffer from having to support processors 30yrs back.
If anyone can suggest just a processor with less historical nonsense, that'd be already great.

[Ethin]

Okay, so thanks for all the answers everyone!

If someone was wondering, the primary reason that I'm trying to find something old isn't even that it's more interesting (although that's very important), but that it doesn't suffer from having to support processors 30yrs back.
If anyone can suggest just a processor with less historical nonsense, that'd be already great.

Try RISC-V. It has some historical stuff, but I like the architecture because its clean and is trivially extensible. Its also super, super small if you want it to be!! Its RISC, though, but that shouldn't be too big of a problem.