Okay, lets examine the differences:
Amithlon. No ECS/OCS/AGA emulation.
UAE/AmigaForever: No AGA
HOWEVER!
There is 180 megs of underlying OS needed for the emulators, on 1,000 plus variants/configurations of HW, AND can true rad: reboots be done?
Emulators also depend on the underlying OSs grumbling stumbling of other activities it does. Boy, you should see my windross xp Pro crawl when I SW disable internet access, and fail to work when I turn off swapping.
The FPGAs, I would call nearly true Amigas, only slightly below ASICs as what they both do is AMIGA and ONLY AMIGA related.
As for cycle exact, the NatAmis should ALL be cycle exact ...... TO EACH OTHER. That is
paramount! That was one of the true purities of Amiga. Generally there was cycle exact across a huge amount of machines.
Of course,
true cycle matching is ONLY when the same list of programs are being run on the same machines doing the same tasks.... well a video game running on the same HW and no other SW multitasking in the background.
Amiga had the unique situation of being a home computer, having an extraordinary USER based OS (
all privileges given to the USER), extraordinary custom HW, and having MOST machines operate identically (relatively) always as Playstaion or Nintendo consoles do to each their own.
The only gripe I have with minimig is the lack of full ram of 2 megs chip, 8 megs fast.
NatAmi60 will prove to be the ULTIMATE COMPUTER* only surpassed by an ASIC running at 2.8+ GHz.
* It's just a fluke that the ultimate computer is an AMIGA.
There is a superior to this version even, and without going multicore SMP. Seems there's no way to go SMP.
Although, how about running AOS3.0 in every core? Separately, four AOSs running at the same time. You could even be running different versions in different custom chipsets. They talk to each other as though they're other computers through fake parallel port links. Just divide up the tasks to different CPUs on your own. Run Pagestream on 1 CPU, a rendering program on another, a paint program in another, etc. A custom memory controller would separate the ram pools so the CPUs don't need to keep track of not overwriting other CPUs data. A fatal crash even couldn't write over another CPUs address space as long as the memory controller absolutely knew which CPU was sending data to ram and never allowing the data to cross ram segments.
How about one core monitoring the others, and if a guru meditation occurs, it can be rebooted from that CPU, automatically? TRY to make a self recovering computer, quadruple uptimes.....
For those who haven't seen it yet,
Still Alive, on YouTube.
ONLY AMIGA makes it
possible!!!!!!!