Alex you know NatAmi stands for Native Amiga right? The following is from the NatAmi website:
It's a re-interpretation rather than re-creation of the Amiga chipset. There are going to be issues.
All features documented in the Commodore Hardware Reference Manual, work as written there.
That's just it. That manual is incomplete. They don't have the original source / schematics. They are reverse engineering the chipset. They are going to miss subtle timing an exceptions not documented. UAE and MiniMig both suffer from these compatibility issues (although as development of both continues information is wriggling out). Toni & Yaqube have done great work tracking these undocumented features over the last few years. They have shared them too which I think is very cool.
The early Natami's will likely have bugs and worse compatibility than other classic Amigas BUT everything can be fixed.
Absolutely. And because MiniMig was open source, when the original developer Dennis van Weeren left the scene, other engineers and enthusiasts were able to jump in and pick it up. The original MiniMig compatibility wasn't great but thanks to Jakub Bednarski (Yaqube) and Tobias Gubener (TobiFlex) and others (Peter Wendrich, MikeJ) it is now pretty good.
Most projects seem to become instant collectors items for the handful of people who could actually beta one before the devs gave up.
Exactly, e.g. You can't really buy MiniMig v1.1 PCB's anymore. But it doesn't matter! People have ported the code to at least four different open FPGA boards. If you know what you're doing you can port it to any future FPGA board (with the right I/O) too. The death of hardware manufacturing isn't the death of open source FPGA designs.
NatAmi being closed source will hopefully not stagnate if the creators enthusiasm slows. Lets see how it works out.
If the N68070 was ever burned into a real chip, it should be competitive with low end PPC (and ARM) in performance while using very little electricity.
Erm how did you come to that? Electricity usage in ASICs is primarily down to the physical implementation. How it was synthesised, partitioned and routed. The logical design has a small part to play but it is a small part. And unless NatAmi thought about power upfront (they may have, but it would be strange to do so for FPGA work) then it is unlikely that they have designed the VHDL/Verilog for ULP.