ARM based Amiga?

[Mrs Beanbag]

Just throwing an idea into the wind here, but with NVidia's Project Denver desktop ARM based CPU threatening to mop the floor with Intel (yeah believe it when you see it), perhaps an ARM implementation of AmigaOS would be a prudent move.  I've waited for so long for x86 to bite the dust, it's kind of amazing how they've managed to keep the 680x0's arch enemy propped up for so long. Plus a surprise victory from the good old Acorn Archimedes, or to my mind the "British Amiga"!

Any thoughts?

[Mrs Beanbag]

@Heiroglyph

Well that would be nice wouldn't it, a 3GHz quad core 68k chip, but where are we going to get one of those?

I'm well aware the Amiga has a past, but what concerns me here is whether it has a future.  I think Operating System support for software emulation seems as much as we can realistically hope for.

[Mrs Beanbag]

"When people think Amiga, they think 68k."

Well I think there is a partial truth in that.  I'd say that would mostly apply to the remaining Amiga fan scene.  To me, the Amiga is 68k too (although more importantly, not x86!).  But if you asked most people in the tech scene, they would say Amiga meant to them the graphics and sound capabilities as well as the responsive multi-tasking operating system and window-based GUI that were ahead of their time.  If you asked most people generally, they'd just remember their favourite games and wouldn't even know what a 68k was.

A modern 68k hardware solution is sadly simply not plausible anymore; but I would agree that any computer calling itself an Amiga should be able to run 68k software seamlessly.

[Mrs Beanbag]

@Boudicca:

"biggest insults" oh come now let's not get petty!  Or would you rather Amiga was beaten to death by IBM's little minions?

Although the irony does not escape me, but it doesn't stop there.  Of course there was also the Apple Macintosh that also ran on 68k, migrated to PPC and managed to survive with a respectable share of the market only finally to admit that they had to migrate again to the dreaded x86.  Now Microsoft has stated that Windows 8 will be available for ARM, and how willing will Apple be to put its customers through another platform shift?  Apple will be out on a limb clinging to x86 while Windows jumps ship to the Neo-Archimedes!

[Mrs Beanbag]

I foresee a wholesale move to ARM over the next few years.  One of the main things that limits CPUs today is the amount of power they consume and hence the amount of heat they put out.  They've been pushed as far as they have mainly because of Intel's clout, Microsoft's software support and the need for backwards compatibility.  ARM, however, requires a fraction of the power.  They're already being pushed for the server market: see http://www.theregister.co.uk/2011/03/14/calxeda_arm_server/

It's the sensible way to go.  68k can only have appeal to old-school enthusiasts; it's much better than x86, granted, but it's still CISC and will still have the same limitations at high speeds.  Besides, nobody programs in ASM anymore.  Even embedded systems programmers now typically code in C.

[Mrs Beanbag]

The reason there aren't more clever one-man assembler developers today is because increasing CPU horsepower has made high-level languages usable even on inefficient compilers, so it's no longer necessary

Well there is that, plus modern optimising compilers pull all kinds of tricks that you'd have to be super-expert to know about.  But have you ever seen x86 asm?  It is the very work of the Devil.

[Mrs Beanbag]

@HenryCase: it does if you put UAE on it!

[Mrs Beanbag]

@HenryCase: yeah, why not?

[Mrs Beanbag]

@HenryCase

I don't know where you're going with this.  It's a Windows tablet if it runs Windows.  It's a Mac tablet if it runs OSX.  It's an Amiga tablet if it runs AmigaOS.  Or maybe you think it needs a 880k floppy drive to qualify?

Windows 7 now needs an emulation layer to run XP software.  Is it still Windows?  Double glazing, maybe.

[Mrs Beanbag]

@HenryCase:

So what's the difference between a Mac and a PC now, just the operating system?  But if you install Windows on a Mac, which you can now do (and vice versa with the right combination of hardware), is it still essentially a Mac?  Hmm... it's beginning to sound like a computer is whatever the manufacturer says it is.

I agree with you in principle but I still don't think we've got to the root of the matter.  What if you build a computer (tablet or otherwise) with generic hardware, and installed some version of UAE that could boot straight up without any intervening operating system?  In other words, UAE in the kernel.  Would that be an Amiga?  What if the hardware was supplied by Amiga Inc?

Or does the difference have something to do with the BIOS, or the ROM?  What if a computer had a custom BIOS that showed an image of a hand holding up a CD when you turned it on?  And you got into the BIOS utility by holding down both mouse buttons.  What if the old EGA text mode were replaced by OCS graphics?

[Mrs Beanbag]

Well I agree with everyone.  I know this sounds paradoxical, because everyone aren't agreeing with each other, but I can see everyone's point...

Here's the way I see it.

Nvidia and others think ARM has a future in the mainstream desktop/laptop market and are putting their money where their mouth is.  Microsoft also agree and when ARM based PCs appear, Windows 8 will run on it.

Apple can migrate to ARM, but unlike Microsoft also build their own hardware and bundle the OS with it as a complete product.  Even though OSX will run on a lot of generic PC hardware you can't just buy it for that purpose, you either get it pre-installed on the hardware or you get an illegal copy.  As such Apple will migrate to ARM for their desktop/PC market when they have to, not just because they can.

So when Denver comes out there will be a gap of several years while Windows has essentially no competition in this corner of the market.

Nvidia are consciously aiming for the low end of the market, not high performance or gaming rigs.  They will be better suited to HTPC.  I see that Amiga has a chance here (if the financial backing were available, of course) to make inroads into the HTPC market at that point.  It won't have the CPU "grunt" but let's face it Amiga never did, it was always a fairly competent CPU backed up by the custom chips.  If you wanted pure grunt you got a PC, even in the Amiga's heyday.  That's not what it was for.  In 1987 even the Archimedes outperformed an Amiga in pure CPU terms, but the Amiga was better for games and other 2D graphics applications because of its custom chips.  As someone else pointed out, "CPU as supervisor with specialised coprocessors to do the grunt work" is also the ARM philosophy!

I think if there is to be any next generation Amiga it will have to be what Amiga always was -  a multimedia PC, and a console with an OS.  There is a desire for that, people were installing Linux on Playstations until Sony put a stop to it for some reason.  You don't need grunt to do that.  Consoles don't have grunt, they have graphics chips (the Xbox 360 actually runs on a PowerPC chip).

And here is how it could work: Amiga as chipset.  Essentially a Natami board with a Denver ARM socket, with the 68N070 coprocessor that monitors/controls the hardware.  Old games will be able to run directly on the chipset (one is reminded of the Sega Megadrive, which had the same Z80 that powered the Master System as a coprocessor, and could run Master System games through it), AGA will replace the old EGA bios text mode, while an onboard (Nvidia?) graphics chip provides gaming performance at Playstation 3 levels.  It will do everything your Amiga does (including running classic Amiga software), it will play games classic and modern, it will function as an HTPC, and it will also function as an entry-level desktop PC if you want it to (web browsing, e-mail etc).

If you want all that and grunt AS WELL, well, wait until the ARMiga 1200 works out, and the ARMiga 4000 might follow.

Are there any objections other than that it's all pie in the sky?  If anyone else happens to have the money, I've got the ideas and the vision!

[Mrs Beanbag]

Oh and for the record, I don't think x86 is evil because it's a monopoly.  I think x86 is evil because it's absurd.  The fact that it is also a monopoly is just rubbing salt into the wound.

I'm going to quote some things from the 1996 edition of Computer Architecture: A Quantitative Approach (appendix D):

"The x86 isn't all that complex - it just doesn't make a lot of sense."

Mike Johnson
Leader of 80x86 Design at AMD,
Microprocessor Report (1994)

Concluding Remarks

As we have seen, "orthogonal" is not a term found in the Intel architectural dictionary.  To fully understand which registers and which addressing modes are available, you need to see the encoding of all addressing modes and sometimes the encoding of the instructions.

Some argue that inelegance of the 80x86 instruction set is unavoidable, the price that must be paid for rampant success by any architecture.  These authors reject that notion.  Obviously no successful architecture can jettison features that were added in previous implementations, and over time some features may be seen as undesirable.  The awkwardness of the 80x86 began at its core with the 8086 instruction set and was exacerbated by the architecturally inconsistent expansions of the 8087, 80286, and 80386.

A counterexample is the IBM 360/370 architecture, which is much older than the 80x86.  It dominates the mainframe market just as the 80x86 dominates the PC market.  Due undoubtedly to a better base and more compatible enhancements, this instruction set makes much more sense than the 80x86 more than 30 years after its first implementation.

For better or worse, Intel had a 16-bit microprocessor years before its competitors' more elegant architectures, and this head start led to the selection of the 8086 as the CPU for the IBM PC.  What it lacks in style is made up in quantity, making the 80x86 beautiful from the right perspective.

The saving grace of the 80x86 is that its architectural components are not too difficult to implement, as Intel has demonstrated by rapidly improving performance of integer programs since 1978.  High floating-point performance is a larger challenge in this architecture.

[Mrs Beanbag]

@Tripitaka: I've been thinking on similar lines for some time in fact, but now I've heard about Denver it seems a good deal more plausible.  We've got a Beagleboard in the house, I wonder if we could get it to play nicely with a natami/minimig somehow...

[Mrs Beanbag]

@Thorham:

Hang on a minute... first you say Amiga is "hardware (680x0 + chipset)", then you complain that the OS is "stuck in the past!"  With all due respect, the 680x0 and the classic Amiga chipset are obsolete hardware.  Don't get me wrong, I love the classic Amiga, but if you insist that that is what Amiga is and always will be, the Amiga is a relic of the past, a museum piece used only by die-hard hobbyists.  I can sympathise, but this is not a way forwards.

In fact I love the 680x0, and I'm about to code some asm in it this afternoon, but it's possibly the least Amiga thing about the Amiga.  Why?  Because it was a generic part.  It was also in Macs and Megadrives and goodness knows what else.  Even the floppy disk drive was more uniquely Amiga (who else got 880k on their floppy disks?).  The OS and the chipset were specifically Amiga products.  Both of these need updating for the 21st century.

@SamuraiCrow: I'm not talking about having the Bealgeboard and Minimig on the same board - perhaps some kind of serial link could be used at first.  A sort of early proof of concept.

[Mrs Beanbag]

@nicholas

The beagleboard has an expansion port. I wonder if a 68k emulator could be installed on it and an interface made to connect it to an A1200's trapdoor expansion port... 600MHz accelerator card for Classic Amigas, anyone?