All programs and data would reside in fastmem

Programs and data can reside in chipmem too.

 

Amiga games are all but AOS software. The problem about accessing hardware registers is an old one and will not be solved without emulation of some kind. However, AOS-compliant software never needed to access the hardware registers (at least starting from AOS 3.0).

There are some things that cannot be done directly with AOS.  For those things directly accessing registers is required.

Some time around 1995 I wrote a small tech demo for a 2D parallax scrolling jump'n'run that was 100% AOS-compliant.

Anyone can achieve quite a lot by using AmigaOS.

Such software would continue to work even if all direct access to the chip registers was barred. And so would all types of productivity software.

Such softwares would only work if
A: The emulation layer was 100% accurate
B: The emulation layer was 100% fast enough

Else you would get glitchy gfx and sfx.

To derail this a bit, I liked that CPU with separate MMUs for instructions and data.

Actually I didn't want to start a discussion about how much we need an MMU. I think that an MMU is a nice thing but I rather have a Natami without one than wait forever. The nice thing about FPGAs is that they are programmable and that we aren't limited to just one design.

I was more interested in starting a discussion how an MMU could have helped making AOS ready for the future (=the time in the mid-90s when the 68k line didn't offer an upgrade path anymore and the custom chips proved to be a dead end). I don't think that much of this would happen in reality but sometimes it is just nice to discuss how things could be done.

And this is with the MMU implemented.

AOS nailed it pretty well and it only part that was lacking which later got corrected with BOOPSI.

The expansion sub system defeats PCI when PnP comes to mind.
the weakness was hot swapping and disconnecting a device at runtime.
A MMU could have helped making it easier for the system developers.