How about creating a Open AMIGA Consortium?

[freqmax]

Call it 'Excellentia' or some other new name..
(Because Amiga in my book was about technical excellence at a economical price)

'Comes comitis'.. 'Duco virtus'..
('Companion friend', 'to calculate excellence')

Or 'BillSucky' or 'BillSuckInk', for those that like the revenge approach :-)

[freqmax]

If I were to make an opensource A1200 (or A4000?). I would need:
 * AGA specification.
 * 68020 specification.
 * Amiga software tailored towards verification of specific hardware functionality. Ordinary games are good, but won't pinpoint problems.
 * Estimation of chip complexity, ie how powerful FPGA is needed.
 * An hardware group to correct schematics, and put it into physical hardware. Multilayer, BGA chips, and highspeed signals requires some serious verification.

As long as we don't need a physical cpu but rather use large FPGAs the last point could be eliminated. However using the real cpu (like Minimig does atm). Is one chip less to replicate properly.
Maybe this will give you an idea what would promote development of new amiga hardware.

I think a danger of formalised organisations is that they tend to get rigid by time. And persons involved might be more occupied to preserve their influence rather than help the overall purpose.

[freqmax]

@persia:
An operating system relies on drivers. So no need holdback. And still have hardware that can act both in game compatible mode. And "modern" mode. This hw flip can be accomplished with FPGAs.

@A6000:
We already have an "common API" be the means of the hardware released by Amiga Inc before going belly-up.
With Classic Amiga being the definition. MiniMig, and CloneA being clones.

Natami, and PowerAmiga (PPC) are extensions. Not neccisarily being compatible in all aspects. But keeping as many as possible aspects of classic Amiga compability.

Amigaone, and Pegasus is something completly else. Only seems to build on the love for motorola style cpus and the Amiga community.

So Amiga compability definition already exist more or less.

Anyone knows how M68060 stands up in comparision to Intel line of cpus?, natiami mentions 300x performance compared to  M68000.

[freqmax]

I should add something else too. For any new software development I would require:
 * Memory protection
 * Preemptive multitasking
 * Multiuser structured security with file permissions etc.

Or else the risk for wrecking data or interrupt running applications is just too great to have a suitable working platform.

[freqmax]

@persia:
ATSC is a transmission and coding standard. NOT a video generator<->display link. Same goes for DVB.
Blu-ray etc.. can be handled via firewire/usb. No need to waste resources in a technological rat-race. I think you should read up on the technology issues.
Using of the shelf components instead of FPGA will limit the capabilities severly. And Amiga is not about mediocrity.

@AeroMan/persia:
I rather get a M68060+FPGA or PPC+FPGA (Virtex..). And then use technical skills to see how much performance one can squeeze out of it. The other way around is like.. we shall be no better than X and Y.

I think the idea to base an software API on the existing kickstart is a good idea to get a well defined API. That way we can easily find non-compliant implementations.

[freqmax]

I no longer invest any resources in anything that can be taken away to will. At minimum the API should be available. And ofcourse even better the source code with a licence to change, redistribute etc.. (BSD/GPL).

Can't these be used as definition?
It will allow existing software to work. It's widely available, and almost anyone can verify compliency. Won't require any major changes as to how things work.

Hardware API definition: A500/A1200/A3000/A4000 etc..

Software API definition: KS 1.3 basic, KS 3.1 full.

Hw/Sw that works as above should be considered "Amiga compliant".

[freqmax]

There's a huge software base for Amiga that I don't want loose. Most newer versions just don't have that software base.
And in any case there's nothing to prevent loading another OS.

If modernity is the way. Then we might aswell dump all Motorola CPU/OCS/ECS/AGA and go for MIPS/ARM/etc.. + Intel graphics (most open atm).

[freqmax]

It might be an idea to have two different tracks for future Amiga development:

(1) A version which is Amiga classic hardware/software true. Ie if you can run it on a A500 it will run without modification.
(MC68060, FPGA chipset, etc..)

(2) A version that is brilliant in using technology. We won't compete with ATI/nVidia/Intel/AMD etc.. but we exploit existing chips in a clever and costeffective way. This version would not have any compability ties with original Amigas at all except in spirit.
(Get the most MIPS/Hz CPU, efficient GFX, FPGA subsystem, etc..)

One could exploit the SRAM feature of FPGAs by reloading it every 10^-3 second. And thus make it do many things with a small configuration.

As for a consortium. Collect data on A500/A1200/A3000/A4000 and put into a report that can be used for hw/sw developers for re-implementations.

An portable Amiga with a 7"-10" screen with super efficient use of battery/cpu/ram/periphials etc.. could be an area where it's possible to get at the competition.

A drawback we as a community have is the economy of volumes, army of optimisation engineers, budget for failed prototypes etc.. any project will have to make up for it in some other way. Once one can produce at least 1000 units. The producer is the king because you can order directly from manufacturers, not digikey/jameco/mouser and the rest.

[freqmax]

@biggun:
It's possible to do without the extra copy stage. It's called "zerocopy" and NetBSD has implemented this. You could try it and see if performance improve.

@abbub:
Modularity is the key to avoid the complexity trap. I also think USB for lowend devices is needed, eventually. Not now because the stopgap solution (PS/2) let's us focus on other issues. In the meantime prices for an implementation will go down.

@Sig999 post 3/26 22:00:
Ethernet can cure all those storage needs. I think the filesystem networking doesn't have the 4 GB limit.

@JetRacer:
 (1) "Amiga was everything-in-a-box. A user streamlined package"
 (2) "each plugin card that added an Amiga feature costed as much as an Amiga"
 (3) "The only reason the Amiga chips exists is because no 3:rd party could provide the technology at the time."

Good points JetRacer!
I think this kind of summarize the economic driving factors behind the Amiga.

I think I should point out from an electronics designers point of view. That adding yet-another-chip to an existing pcb is very cheap unlike doing the same with an external card.
 
So an hw/sw true box would be an FPGA+PPC with a PCI slot I guess. And the amiga spirit compatible would be efficient & modern asic cpu + fpga.