That does bring up a good point. It would be nice if the original system could be productively used as much as possible.
@TheRouge:
Given that the Zorro buses on the old machines are locked at a certain Mhz, is there any way around the bandwidth limitation by on-the-fly compression and decompression of the bits sent over the bus?
Say a new accelerator is introduced and it want's to communicate with a new RTG card on the old Zorro III bus. Would it be possible for both the new accelerator and the new RTG card to have a chip which very quickly encodes and decodes data, in effect increasing the amount of information carried over the bus per-bit? Perhaps something like this would allow the bus to handle more traffic?
Sorry if this is elementary or the way things already work. I'm sure someone else has already done something like this, but I'm not sure if its been applied to the older Amigas...
From my understanding of Zorro III, this would not be possible. Communication between the CPU card and something on the Zorro bus would have to be relayed through the Buster, which has a practical limitation of ~20MBps. From my understanding of things (While I do have experience with hardware development, I have never attempted an Amiga accelerator before) the only way you could boost the speed of Zorro communication would be to design a replacement for the Buster chip. While this is theoretically possible to do using an FPGA and a plug-in adapter, I'm not going to commit to something like that without a real good reason.
That block diagram seems to almost describe a complete computer. Why even bother with the Amiga part?
I've always asked myself the question: Why? Add a graphics card and a sound card, and you've got a PC with a 680x0 CPU. Not cool
Keep it Amiga, guys 
The point would be to allow more demanding software to run on the Amiga platform while keeping close to full backwards compatibility with older software and hardware. I would not consider high-end graphics, large amounts of memory, some PCI-e lanes, and a fast processor attached to an OCS/ECS/AGA chipset to be a PC or close to it. I consider it to be an Amiga with high-end graphics and sound capabilities and an upgrade path (4 PCI-e 1x lanes) that is more future-proof then what is currently available. By your definition a CSPPC+CVPPC+128MB is a complete computer as well and therefor pointless. I'm not trying to be antagonistic or anything, you are totally entitled to your opinion and I can absolutely see your point of view, I just think that as long as you are running a real 680X0 on a machine with a real Amiga custom chipset, it's still an Amiga, the rest is just gravy. You can still put an Amiga floppy disk into the drive and boot an Amiga game without any sort of emulation taking place. I actually find the idea that OS4 Classic uses JIT instead of the real 060 makes me unhappy. I fully understand that it was done for a good reason and I am by no means criticizing the developers, I'm just making a point. I know some purists would say a design like what has been proposed in this thread is absolute heresy, but I also think a lot of people will be very happy to have it.
Dont worry are just fantasies, people starts with accelerator cards and ends with another more "amiga friendly" system.
I absolutely understand your skepticism and I take no offence from it. I hope you will not take offence when I say that I look forward to proving you wrong.
@TheRogue
Holy Crap!(tm) If you can make something like your block diagram that is just
Freaking Awesome!
About your "expansion area". What is easier for you hardware guys to add these days? An oldsk00l PCI slot or a newfangled PCI Express x1 slot?
I bought a Gb Ethernet card 1000/100/10 for a PCI Express x1 slot the other day for like $40.00. I did not check if they make combo cards with Gb Ethernet and USB all in 1 card.
In any case if the card had some sort of way to plug in an off the shelf Gb Ethernet and off the shelf USB solution then that would be really really great and would solve a whole lot of problems all in 1 go.
I am not personally a fan of USB 1.1 to 2.0. I can't stand it really But I respect that other people need it for stuff. But I do love USB 3.0
Thank you for your enthusiasm! To answer your question, as long as your hardware can support the speed PCI-e is easier due to it's serial nature. PCI is a parallel bus which means more traces, more I/O pins tied up, etc. I think it would be best to implement a header with 4 1x PCI-e signalling lines that could either be connected to a bus-board for connecting full size PCI-e cards for a tower configuration, or small board that plugs directly to the header and passes through the remaining PCI-e lanes to another header for use in tighter cases. It would obviously be smart to implement the newest version of the USB standard instead of the older versions at least in my opinion.
