AAA machine today?

[Marchie]

Just throwing out a crazy idea - would it be possible to manufacture a AAA board today?

I've seen the Youtube vids of the non-working prototype with all the schematics and am just wondering in these times when we can get custom PCBs made, would it be possible to 'print' the PCB and implement all the custom chips in FPGAs?

Would that not be the coolest ever classic Amiga?

Don't know what I'm suggesting here on a technical level - whether the information exists to do it, how much work it would be, (or whether I'm the only one who thinks the idea is worth doing at all).

Just wondering since I was under the understanding that the AAA system actually did work (as opposed to the Hombre chipset which was never actually completed)

Chime in engineers! (if only Dave Haynie read these forums!)

[grelbfarlk]

Not much point to it. Let's look at it this way, what is the advantage to say an AGA board over an ECS board? AGA software (basically games.) AGA software runs on AGA and not on ECS basically and a little more performance. There is no software written for AAA. RTG was the solution to this problem, you write for RTG then you aren't bound to whatever OCS, ECS, AGA chipset you have on the motherboard, similarly with AHI.

If you're suggesting ok lets just make a new Amiga motherboard with faster stuff, I think that's a fine idea. Wider and faster buses and a standard interface to GFX and Sound whether that be PCI, PCIe, AGP, Thunderbolt, Firewire or USB3. You still have to shoehorn some kind of 68k onto there or emulate in FPGA along with an Amiga chipset for backwards compatibility.

At some point you start designing this and say what is exactly the point to build a new motherboard for the low volumes of the Amiga market and also low performance with a high price before you start to wonder what's the point of it all, why not just take an x86 board and put your Amiga chipset and 68k on a PCIe card and be done with it. Or no PCIe card at all and it just boots straight to AmigaOS like Amithlon.

Or you want to be compatible with NG Amigas too so now you're doing all of the work of the 68k and FPGA but also building a PPC motherboard. So you get pretty fast PPC main CPU, and even higher prices.

Countdown before someone mentions FPGA based standalone Amigaish compatible motherboard is right around the corner that is pretty slow and pretty expensive compared to all of the other solutions but faster compared to classic 68k. But it's totally an Amiga-for real-seriously.

grelbfarlk said it better than I could. Dave Haynie himself did talk about it once too; I think it was a YouTube video (I'll have to see if I can find it again).

Long story short (and paraphrasing from what I remember he said): Although the AAA chipset was in development, it was far from finished. You could theoretically put the work into getting it into a shippable product, but it's so far behind hardware and software wise compared to what's out there that it's pretty much not worth it. Then there's the whole issue of what would you run on it?

Mind you nothing is impossible. The closest project that I can think of that compares is the Commodore c65 project: http://c65gs.blogspot.com.au/ Obviously it's not a question of whether it can be done, it's more of a question of whether it'd be worth it in the end.

[Djole501]

Vampire standalone board (and the current v2 accelerators perhaps) will have some of the AAA features, like 16bit sound, improved AGA and so on. Time will tell if software for this will be written.

[Marchie]

I think you're right grelbfarlk, there are no features to be had from the AAA board that we don't have 10x better in modern systems. And the point about hardware without software to make use of it is hard to dispute.

I guess the AAA board seems like the last and best genuine Commodore Amiga computer, that was so close but never quite finished, and I'm just wondering whether time and technology have enabled us all to complete that Amiga A5000 and see what it would have been like.

Would it be worth it in the end? Any normal person would say no, but an Amiga enthusiast might (might) just say yes.

[grelbfarlk]

Since it's not possible to find faster Lisas, Alices, Garys, Ramseys, Paulas, CIAs, Busters and certainly not cheap, you're looking at putting all that stuff in an FPGA. At that point do you define a new Accelerator slot or try to be compatible with an A3000/A4000 slot? I'd say define new. Here's my take
1) Standard motherboard format ATX, mATX, Mini-ITX something like that.
2) Assuming the machine can boot with just the motherboard or with a processor card and has Video, Sound, IO
3) Need to decide on an expansion bus, do you go with busboard design leaving open the possibility for Legacy Zorro cards with PCI or PCIe? PCIe would be nice but drivers are going to be the death of you, but lets say that's not a problem-go with PCIe. Or go with integrating the Prometheus on the motherboard so you have at least the small ecosystem of OpenPCI drivers to start with.
4) Lets go with a new CPU slot design with fast bus to the PCI/PCIe, so you could in theory slap a 68k CPU card or even a PowerPC CPU card in for OS4 support-or a card with both.
5) All your crappy old IO goes here (that I personally prefer) 2xDB9 Joystick/Mouse ports, Serial port, maybe even Parallel, with some USB at least HID ports. No more 23-pin RGB, just use a VGA 15-pin or even Displayport-I guess both so CRT lovers don't complain.

So that's it, give me a million dollars and I'll have someone build it.

But at some point you look at this thing and it looks a lot like an NG Amiga, with the exception that you have the Amiga chipset on an FPGA on the motherboard. But you kept Zorro slots (insert reason for this here).

Why not just put the Amiga Chipset+68k on a PCIe or PCI card and slap that in one of the NG Amigas? Or just in the slot of your Wintel machine. Then what did you create? A slower way to run WinUAE. And maybe in the FPGA card is in the NG Amiga a little faster way to run FS-UAE (I don't know how fast JIT on PPC is maybe it's still slow).
That should be relatively cheap at least-but still rather pointless.

Maybe they aren't capable of booting off of a PCI card or it's too counter-intuitive-or they are already too expensive and having the fake Amiga chipset on a PCI card isn't real enough for the 68k purists.
Or really the main reason, what's the point of any of it, since that FPGA isn't going to be nearly as fast as the main CPU in the NG Amiga anyway, certainly not the GFX or the rest of it. At least you can say what's better than one Amiga? DOUBLE AMIGAS IN THE SAME CASE.

But lets say you could just boot off the Amiga on a PCI card and you put a great big switch on the front of the Amiga to boot to classic and there you have AmigaOS 3.9, then whatever host CPU is sitting on the motherboard is there for 3.9 to use like WarpOS.

In any case all that's going to cost some bucks and take years to do and be pretty expensive.

Or lets just forget all of this and start over fresh with a new Amiga Operating system on fast hardware-which is what AROS and MorphOS and others are doing-yes I realize they sort of run on classic hardware too. Or just port AmigaOS 4.x to x86. Or since the AmigaOS 3.1 source is available just port that to x86.

Knowing the Amiga community I expect all options will happen in opposition to each other.

[Marchie]

Originally Posted by grelbfarlk

Knowing the Amiga community I expect all options will happen in opposition to each other.

True that.