Who currently owns the rights

[minator]

I'm not talking about a new Amiga, I'm talking about a scaled down Amiga Retro Games machine. The rights to games are easy to aquire - all you need to do is sign papers and pay the money. Any company that sees that they could potentially make a few grand or more off a title that has made them $0 in the past ten years is more than happy to accept your check.

It could actually work, the chips could probably be combined on a single chip these days - including the CPU.  It wouldn't need cutting edge chip technology so it'd be cheap to develop as well.  You could probably do it as a hybrid using the main custom chips and a better CPU to emulate the other chips and CPU.

If you didn't want to develop a chip (which is still a fairly serious investment) it could be done using a special purpose PC.

But alas! There hasn't been a company with smart business sense in the Amiga technologies realm in the past 15 years (Commodore included).

They'll never go for it, it's a sensible way for Amiga to make money and that'll never do...

[minator]

the technology can't even be FAB'ed any more

Any modern fab should easily be able to manufacture the Amiga chipset, they would need modified for the different silicon process but that's nothing spectacular.

The problem with Amiga custom chip tech is that you can't simply miniturise it and put it all on one chip, the chip designs simply won't work anymore.

You wouldn't "just shrink" them (doesn't work over more than a few generations), they would need to be re-synthesized but they are simple chips by todays standards so that'd hardly be a problem.

The most difficult bit is probably getting the original CAD models and finding something which can handle them today.

If the gate level designs still exist in a CAD package there's plenty of companies who can produce a semi-custom "Amiga on a chip" for the right price (with optional 24bit Gfx and 16 bit sound).


I actually think it's a good business proposition.

[minator]

If you want to do it, then the best idea is probably to build the whole thing (CPU and all) in an FPGA, with a simple RF tuner + some DACs + a rom (witha kickstart + a simple workbench and ADF loader). Then have a CF or SD slot for ADFs.

FPGA is one way but there are also psudo-custom chips which are existing systems with a customisable block.

Another way might be to use one of the new multicore embedded CPUs (20+ cores per chip), it takes a lot of power to do a good emulation of the chip set so you could break up UAE and run bits of it on each core, it may then be fast enough to run it in software.

Multi-core may not work that well though and requires additional components anyway - and has GPL issues...

Wouldn't be cheap though.

If you wanted to produce a system like this it's going to be expensive anyway, you can't make them cheap to sell unless you produce them in fairly serious volumes and we're talking $2 million + for a 100K unit production run (assuming it's $20 to produce).

Even if you spent $2 millon on chip development that's another $20 per unit but that assumes this would be the only production run.

I think the best way wold be to take an existing System on Chip and modify it to include the Amiga chip set (or parts thereof) as a module, this'll produce modern device with Amiga compatibility.

It's no small undertaking and definately not cheap but there's no reason that it can't be done.

[minator]

I don't see what this argument is about, there is no fundamental reason that this can't be done.  Yes there are problems to solve but nothing that *can't* be solved.  It wont be easy or cheap but they're not reasons to not try.

Regarding IP you seem to be going by different definitions one says "owership rights" the other seems to think it's specific designs.

Both are right but the real problem is can the original designs be reworked into a form which can be used with a modern IC design process?  Again not cheap but almost certainly, and even if they can't UAE and other docs can be used.

[minator]

Maybe the SH-4 isn't perfect, but UAE is mostly CPU intensive... with careful optimiseation UAE could easilly get A500 speeds... though I'm not sure about sound quality...

I've heard more than one report that even a high end PC can't fuly handle UAE (in AGA mode IIRC) so I'm not fully convinced about a CPU based emulator, especially on somthing with as little computing power as an ARM.  Something with a "lockable" cache should do a better job though.  A lockable cache allows you to map all or part of it to a specific area of RAM and use it as a high speed on-chip memory.

If you read Dave Haynies post, this says that it's not much. And if you were recreating the functionality of such chips now you could probably get the whole lot on to one FPGA.

Yes, they go to very high densities these days.
In fact the last Amiga custom chip "Toni" was an FPGA.

--

I'm wondering if there is a market beyond just a $25 toy (which is essentially what we're talking about).

I'm talking about a higher speced machine costing anything up to $100 and sold under the banner of a "retro computer".  The people interesed will be grown up with jobs now so the price isn't a big deal, if sufficiently powerful it could emulate a number of different computers so the market wouldn't just be previous Amiga owners.