The Unitron 1024 Revealed! Can we emulate it?

[eharmon]

A fair bit of information has been spilled about the Unitron 512, a Brazilian clone of the 512K (here's Low End Mac's take: https://lowendmac.com/2016/unitron-mac-512-a-contraband-mac-512k-from-brazil/).

Less, however, has been mentioned about the Unitron 1024. Unlike the 512, which seems like a near exact copy of the 512K ROM, the 1024's ROM has been extensively modified. If it's a fake, it's a great one. There's changes from top to bottom, though it seems to also have been derived from the 512K ROM and is more like a 1024K than a Plus. My first guess is that it's real.

I plan a more detailed writeup on my findings soon, but I wanted to enlist community help first: has anyone successfully emulated the 1024?

I suspect the answer is no. I tried Mini vMac, and it unsurprisingly dislikes the checksums. PCE accepts them but boots to a black screen. MAME claims to have Unitron 1024 support, but its checksums are actually for the Plus ROMs. You can tell it to ignore those, but it gets stuck on a white screen.

But there is one system that does seem to work, a MiSTer FGPA running a derivative of @bigmessowires's Plus Too core, configured as a Plus with 1MB of memory. That takes me into the ROM boot screens where we already find a few differences.

The "happy Mac":


Insert disk:



And the "sad Mac":


It's not just simple icon changes though...it's a rather distinct ROM.

Fun to explore! Except one problem...no OS seems to work with it, thus my extensive experience with the "sad Mac" thumb. It seems to have Hard Disk 20 support patched in...so maybe it expects to boot from there? Unfortunately this Plus Too fork doesn't offer support for that.

Anyway, I think this is pretty neat! I've not seen the 1024's software experience documented well...anywhere!

Any ideas? Incomplete FPGA support, or do we need to find Unitron software too?

 

[eharmon]

Turns out I'm wrong. After a LOT more fiddling with PCE I got it to boot Finder 1.1, strange mouse icon and all!


Why MiSTer won't boot is a....mystery.

 

[demik]

Holly shit. Hope this get somewhere. Nice work.

Is there a ROM dump somewhere so we can compare with @bigmessowires ROM disassembly ?

 

[adespoton]

If you want to get it running in Mini vMac, one of the compile-time flags allows you to disable checksum verification; I use a variant with this flag set to do anything with hacked ROMs. https://www.gryphel.com/c/minivmac/options.html#option_chr

 

[adespoton]

Might be worth checking with @Arbee as well; there may be specific things you need to do to get it booting in MAME (I haven't tried yet). ROMs are available here: https://github.com/DosFox1/Unitron-1024-ROMs

 

[eharmon]

If you want to get it running in Mini vMac, one of the compile-time flags allows you to disable checksum verification; I use a variant with this flag set to do anything with hacked ROMs. https://www.gryphel.com/c/minivmac/options.html#option_chr

Ah thanks for the reference! Looks like it might require some source modifications as well, as I'm booting it in PCE as a 512K with 1MB memory (since the ROM looks more like an expanded 512K ROM than a Plus ROM). Roughly, that appears to be enough to get things going if anyone else wants to give it a try.

I should also compile PCE natively instead of running it through wine...

So far it appears likely this ROM can only boot from 400KiB disks, unsurprising due to its 512K pedigree, but annoying. It will mount 800KiB disks once booted with the PCE ROM extension (Sony driver) + booting into a System 4-based disk. Just like a 512K, it won't boot System 5.

 

[eharmon]

Some more notes from today:

• At least in PCE it refuses to recognize any HFS disks, meaning only 400k disks actually work (I was wrong before).
• It's definitely less stable than a real 512K ROM.
• But it does seem to support up to System 4.0/Finder 5.4 just fine. I tried many old shareware titles and they seem to function.
• System 4.1, which supposedly works on a real 512K, also works if you move it to a 400K disk but starts to show slight issues, like the "safe to turn off" dialog not displaying.
• Hacked System 4.2 and 4.3, which work cleanly on an (emulated) 512K once the ROM patches are copied and some fonts and things are removed to fit on 400k back also work. But they exhibit very strange bugs, with fonts rendering improperly and a white box on screen causing a lot of problems. Clearly at this point the lack of correctly applying ROM patches starts to have ill effects.
• The ROM supports various memory configurations (I tried 512k and 1024k), but only up to 1024k. Interestingly this means a 512K with 1024k of RAM hacked in might also be able to use this ROM.
• I haven't been able to test the possible native HD 20 support, as I'm not sure any emulators ever offered it?

So I'd say while being a very hacked up ROM, it is relatively compatible with everything up to System 4.0.


I'd love to try mini vMac too but I'm having problems with the copies I build myself failing to boot anything.

If you want to try it yourself, the PCE config is...well pretty simple:

1. Grab the Unitron ROM.
2. Open PCE's Plus configuration file and point it at your ROM.
3. Set the memory size to "1024K".

Obviously, it's not a pure Unitron hardware emulation, but presumably it's a true clone so everything seems compatible. Why I had so much trouble to start I can't remember.

 

[eharmon]

Seems DosFox, who re-shared the ROMs, is way ahead!

I found this awesome thread, I highly recommend checking it out: https://tech.lgbt/@DosFox/114053665154970955

I've seen more than a few references to Unitron boards being Plus clones, including DosFox build, which makes things even stranger.

Given their success with the HD 20 init for HFS support, I poked around a little more there. It seems the data fork of the System file from the HD 20 Startup disk is embedded towards the end of the Unitron ROM...which doesn't have the effect of making it work without an init, at any rate.

Unfortunately PCE's support for floppies seems to rely on a startup callback from their driver extension, and things go off the rails once the updated driver is loaded, so HFS support remains elusive in emulation.

 

[nathall]

• Hacked System 4.2 and 4.3, which work cleanly on an (emulated) 512K once the ROM patches are copied and some fonts and things are removed to fit on 400k back also work. But they exhibit very strange bugs, with fonts rendering improperly and a white box on screen causing a lot of problems. Clearly at this point the lack of correctly applying ROM patches starts to have ill effects.

I’d imagine one would need to create a custom set of patches tailored to the slightly different ROM in order to make System 5+ work properly.

The choice to replace the mouse pointer with that awful hand is bizarre. It seems like a strange and arbitrary change, definitely not one made on trying to avoid legal issues, considering the blatant trademark infringement happening with the rest of the physical product.

 

[eharmon]

I’d imagine one would need to create a custom set of patches tailored to the slightly different ROM in order to make System 5+ work properly.

The choice to replace the mouse pointer with that awful hand is bizarre. It seems like a strange and arbitrary change, definitely not one made on trying to avoid legal issues, considering the blatant trademark infringement happening with the rest of the physical product.

It definitely seems like a much stronger attempt to make it look like they "clean room" engineered the ROM since it has totally unique assets. I think that's also triggering the bug on 4.2+ (5.0+). It seems the "Welcome to Macintosh" screen interferes with their custom splash screen and refuses to dismiss properly.

But a simple look through the ROM listings shows you it's heavily copied from the 512K ROM. I wonder if this is a beta version, one where they're half-way through rewriting everything. But it sure looks more like illegal cloning via "dirty" reverse engineering more than a "clean room" implementation. That said, the amount of surgery done is technically impressive.


HFS and 800k floppy support is now working for me. I switched to PCE's direct IWM emulation instead of their custom Sony driver, and moved to the HD 20 init from the v1.1 disk. Copying that to a System 4.x disk works properly. I'm booting off 800k MFS and reading 800k HFS disks. Seems that's the magic touch for real 512Ks too, I've just never had to do it myself.

 

[nathall]

It definitely seems like a much stronger attempt to make it look like they "clean room" engineered the ROM since it has totally unique assets. I think that's also triggering the bug on 4.2+ (5.0+). It seems the "Welcome to Macintosh" screen interferes with their custom splash screen and refuses to dismiss properly.

But a simple look through the ROM listings shows you it's heavily copied from the 512K ROM. I wonder if this is a beta version, one where they're half-way through rewriting everything. But it sure looks more like illegal cloning via "dirty" reverse engineering more than a "clean room" implementation. That said, the amount of surgery done is technically impressive.

What little I’ve read on it, which is probably the same as you, it sounds like dirty reverse engineering is exactly what they did. Supposedly it was mostly coded in C and was 128k instead of 64k, despite being a clone of the real 64k ROM and not the 128k.

HFS and 800k floppy support is now working for me. I switched to PCE's direct IWM emulation instead of their custom Sony driver, and moved to the HD 20 init from the v1.1 disk. Copying that to a System 4.x disk works properly. I'm booting off 800k MFS and reading 800k HFS disks. Seems that's the magic touch for real 512Ks too, I've just never had to do it myself.

That makes sense. I’ve never tried the 800k MFS trick on my 512k since learning of it, recently. I created 400k boot disks years ago for Systems all the way up to 6.0.3, pruned to fit by removing unneeded fonts, DAs and resources from the System file (color assets, etc) and keeping just a couple Control Panels. That HD20/HFS driver takes up 39k or so, which is expensive on a 400k disk. You could do a pretty decent System 6 boot disk for the 512k with 800k, one of these days I’ll play around with it.

 

[adespoton]

Ah; you guys got ahead of me; I was going to suggest copying the init over from v1.1 for the HD 20. Glad to hear it's progressing well! Interesting that the ROM appears to already have that embedded; I wonder if they just never got around to patching it correctly?

And yes, IWM emulation is the way to go. You should be able to use MFS on everything up to a 20MB image though, and after bootstrapping HFS, you're good up until way beyond what the memory can handle. All you should need to put on a 400k disk is enough to bootstrap the HD20, and then everything else goes on a 20MB image.