ARM based Amiga?

[ChuckT]

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?

I think one of the problems is "Who is going to build it?"  If you paid the Natami team as much as engineers get paid, how much money would you need to raise?  If you were to pay a team of engineers to put the Amiga on ARM, I would welcome it but how much would you need to raise to pay some engineers?

I remember from the 80's that some things worked well on the Amiga and other things worked well on the MAC or the IBM.  If you are porting things to ARM then they really need to be redesigned to be hardware specific and that might mean re-writing the Amiga Operating System.

I can't speak for Hyperion but supporting multiple platforms may or may not be worth it to them.  And if some things work better on different hardware, would everyone be happy?

There are different people here who already are heavily into their niche that I bet most won't support moving to ARM due to cost, they are already invested in another system or they have enough they are involved in.  It was the same boat I was in as I put all my money into Amiga instead of Eprom burners, the required books on different chips, surface mount soldering irons and equipment.  The Do It Yourself Movement with computers went broke in the 80's because normal people couldn't keep up with the big companies because it cost too much.

There are a handful of home made or private companies that make their own computers these days but there already isn't enough support.  I think that Amiga.org should consider an electronics forum should there be ARM, microcontroller support, FPGA experimentation or other stuff and then you could recruit some users to get involved but I think you need to reinvent Amiga and be satisfied with what you can do with the current available hardware out there.  Because there won't be support and because we can't compete then there will be SAMs, AROS, emulation, etc.  You can't really expect non-engineers in this forum to learn geek stuff written by engineers for engineers so that they can make an Amiga like computer, can you?

[Tripitaka]

How we think about ARM has an awful lot to do with what one thinks of as "An Amiga" in the first place as well as about future hopes.

To my mind the Amiga was more than just an OS. It was about fun and accessibility, knowing what the files are on your drives, and about knowing where to find them. It was also about clever and elegant design from PCB upwards. The CPU was a huge part of this and the custom chips too. They didn't just provide processing power, they gave us a level of compatibility that PCs didn't and couldn't match. I knew every game I bought for my 'miggy would run as long as it said AGA on it (my first 'miggy was the brand new 1200), it was as easy as a console. All of these things played a part in my Amiga experience and many other things too, being able to plug the sound directly into my HI-FI for games of Xenon 2 and the video compatibility for example.

Well, now is another age. So given complete choice over the direction Amiga should take I would make Natami official, make an add-on ARM board for OS4 and also develop a more powerful ARM based machine incorporating the Natami chips.

Denver plus Natami sounds like a good start for a wedge with a cpu expansion slot that could take a board with a second ARM and GPU combo (another denver )to run in parallel with the motherboard.

Well I know that would cause some issues but that's the only two cents I'm giving today because I'm off to bed. G'night Amigoids! I'll come back to this thread tommorow as it's getting interesting.

[amigadave]

Never happen, but a table Amiga would be awesome.

Probably a "typo", but I agree that an Amiga computer built into a glass topped table would be much more interesting than an "Amiga tablet" computer.

Tablets are so over-hyped!

[fishy_fiz]

Sure, its maybe not as elegant at heart as 68k or even ARM, but when I say "nicely designed" I wasnt really talking about instruction sets, but rather the manufacturing perspective. And x86 of today is a far cry from x86 from yesteryear anyway.

And as you mentioned development software takes a lot of the "quirks" away from a coder (except asm, but that's far from a prominent language these days). Theory and practice are pretty different things. (ie. asm will in some places be faster, but not many people code in asm).

Now in regards to raw power, I think a lot of people disregard hpw useful it can be. Necessary, no, but noticable, absolutely. While its true that for Joe Public raw performance for things like Web Browsing, watching videos, using facebook, etc. is sufficient, it doesnt stop the fact that it is a nicer eexperience, even for those tasks when more power is available. Multitasking for one thng. Joe Public is inclined to "double click this, double click that, do this while thats loading", and so on. Better performance obviosly helps things here. Things like compressing/decompressing files happens often on a computer, and that is much quicker with faster hardware.
Graphics are popular these days. Oviously 3d is much quicker with faster hardware, but even filters for gfx apps (photoshop/paint shop pro/gimp/etc.), not to mention things like flash. All can benefit massively.

It's not so much that its always necessary,but rather its nice to have for when it is.
The Amiga, with its legacy of being great for media centric things isnt a machine well suited to "getting by". Yes the OS runs nicely on lower spec gear, but Id rather all software I want to run has the potential for running in a way Im happy with rather than using a less friendly system, just because its hardware is better suited to the job.

If amiga was a games console, or a netbook, or a slate device, etc. then Id be happy making do, but its not. It's a great desktop OS for being creative with, which requires raw power at times.

All in my opinion/from my perspective of course.

@mechy

There's a good chance youre not actually interested in "truth" here about AROS, but you was so far off the mark I cant refrain....
AROS has nothing to do with amiga you say? How about the fact it runs on 68k Amigas and runs 68k amiga software without emulation? (does OS4?). It also runs on Sam boards. It uses gfx.lib and cgx. Is api compatible with os3.x (compatible, not restricted to as some ppl like to throw around). Uses poseidon as its USB stack. Uses effectively amitcp (albiet upgraded) for tcp/ip stack? Software written for os3.x needs to simply be recompiled to run (you know, like OS4 unless its using emulation for 68k (like aros does when not running on 68k cpu)). Uses MUI/Zune. It's pretty fair to say it has as much to do with amiga as os4.x, the only thing missing is the trademark. To say a system whose purpose for being created was as a compatible re-implementation of amiga os is not just naive, its somewhat bizarre. Its pretty much like saying ReactOS has nothing to do with Windows, or Zeta having nothing to do with BeOS.

[commodorejohn]

Sure, its maybe not as elegant at heart as 68k or even ARM, but when I say "nicely designed" I wasnt really talking about instruction sets, but rather the manufacturing perspective.

Well, sure, Intel has the best fabs and processes, but that's not a virtue of x86, that's a virtue of Intel. Any architecture would benefit from their technical capability if it were available. But conversely, using all that manufacturing might to prop up an architecture like x86 is kind of a waste of potential.

(Especially since they don't even do x86 in hardware anymore, the instruction set is all microcode on proprietary RISC sub-architectures anyway, since that was the only way they could get such a kludge of a CISC system to run at decent speeds - and that was all the way back in the Pentium Pro era!)

Anyway, my point wasn't that more computing power is bad or frivolous, it's that I don't see why Amigoid systems should always have to be cutting-edge (or, as happens in practice, trailing-edge.) If I need raw power it's simple enough for me to drop a few hundred bucks on a no-frills i7 system; for pleasure computing (i.e. Amiga,) I'd much rather have something that's elegant from the ground up.

[fishy_fiz]

Heh, again I dont specifically mean the fab processes, more just that the progress of x86 is a well oiled machine. PLentiful, cheap, powerful and with nothing really stopping it being the cpu of choice for a desktop system. Even power consumption and heat disappation are very good these days, especially when one considers the raw power packed inside.

Given the choice of somethig 10x as powerful for the same price, whose main downside is an instruction set that I very rarely deal with directly (ironically as x86 became more risc like ppc became more cisc like for similar, albiet opposiing reasons), I'd take that any day over the weaker option with an advantage I seldom directly use.

Sure, AmigaOS runs ok on low powered gear, but give it something even trailing edge and it really shines. Perhaps ironically though I probably favor ARM vs PPC for the sole reason that its a lot more practical (cheap/available/etc) than ppc. X86 to me though makes the most sense given the limited resources available to the amiga world. Surely it makes more sense to utilise what is readily available than to spend time and money developing niche systems that offer no advantages?  To each thier own I guess, and this is all purely from a hypothetical perspective.

I personally favor no-one or nothing but my personal experience.

[commodorejohn]

Heh, again I dont specifically mean the fab processes, more just that the progress of x86 is a well oiled machine. PLentiful, cheap, powerful and with nothing really stopping it being the cpu of choice for a desktop system.

Well, yeah, with the force of the entire PC industry behind it, it has to be...

Surely it makes more sense to utilise what is readily available than to spend time and money developing niche systems that offer no advantages?

Plenty of people already doing that, though. Heck, x86 is one of the few general platforms AROS runs native on (general as opposed to specific boards, that is.) Niche systems might not offer any advantages, but they can be fun.

[vidarh]

Heh, again I dont specifically mean the fab processes, more just that the progress of x86 is a well oiled machine. PLentiful, cheap, powerful and with nothing really stopping it being the cpu of choice for a desktop system.

With the rise of the power of smart phones and tablets, I'd argue more people will use ARMs as the CPU of their main computing device in less than 10 years, probably less than 5.

ARM already outsells x86 CPU's by a massive factor. ARM licensees are slated to ship something on the order of 5 *billion* CPU's this year, and even though smartphones and tablets is a small minority of that, combined they are likely to outsell PC's if not this year then next year.

Even PPC and MIPS are contenders in terms of volume (hundreds of millions sold - MIPS licensed 500 million units last year) - they're just focused on other niches, and can't touch ARM even though MIPS has a stated aim of going after the Android market.

This translates into a massive investment in all these three architectures, in growing markets, while the niches x86 sells best in are at best stagnating.

With more and more phones and tablets becoming powerful enough to be comparable to the low end desktops and laptops that make up the vast bulk of the x86 market, and more and more of them sporting HDMI out and support for bluetooth keyboards and support for external storage (even my crappy, cheapest possible Android pad has support for being a host device for external USB storage), and people increasingly spending more money on their phone than their laptop or desktop (my phone outside of contract costs about 20% more than my laptop...), the traditional laptops and desktops are set for a sharp decline, and with it x86 is set for massively increased competition.

That's not to say that it's demise is in any way imminent, but x86 is already a niche market making up a small sliver of the CPU market, and has been for years.

It's just that it's an incredibly profitable niche and very in our faces because we've been so tied to our desktops and laptops.

[Turambar]

And while ARM doesn't seem to come in a desktop board (that I'm aware of anyway, perhaps I'm just ignorant), we could have them with similar difficulty/expense as we get our PPCs in.

I think a http://pandaboard.org/ would make a perfectly capable desktop machine for most everyday tasks.

[fishy_fiz]

@Vidarh

None of which makes ARM a great choice for a desktop. The current mobile fad will subside in time. Already things are close to saturated and more and more people are getting tired of the fad all the time. This isnt to say it'll vanish, but there's not much more scope for extra customers. The market is already predominantely made up of trend followers and same people upgrading thier phones, etc.

The cheapest and most powerful (by a huuuge margin currently,.... despite what people like to claim, arm is a far, far way away from even the most budget of x86 cpus. Even Denver is, when being optimistic, aiming for P4 class performance, something I personally couldnt deal with) will always be best suited to a desktop machine. Currently that's x86.

[tone007]

After 3 weeks using a laptop with a 4-core i7 processor and enjoying the speed with which everything happens, I can't see such low power processors having an advantage where a real computer is needed.  Tablets, phones, sure, they need the battery life, but waiting for web pages to finish rendering due to wimpy processors is no fun.

[takemehomegrandma]

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'm all for it ("AmigaOS" for me being "MorphOS")!

[Hattig]

I agree that ARM would be a good platform for AmigaOS to run upon.

And many ARM SoCs are very Amiga-like in their methodology, having dedicated co-processors for specific functions (video decode, video encode, security acceleration, graphics, audio, etc) besides the ARM core(s).

In addition they are fairly cheap for even a fairly powerful SoC. They aren't on quad-core Intel i7 territory, but they're not half bad.

The Raspberry Pi will be a $35 computer (with ethernet, etc, the $25 variant is rather too limiting). However it is only an 800MHz ARM11 - two generations behind the ARM Cortex A9 used within something like the Panda board. The A9 is competitive on a clock by clock basis with Intel's Atom cores.

And at some point there will be quad-A9s (NVIDIA Kal-El, and others), and then A15s at 2.5GHz. By then we are at the 'enough CPU power' stage for a majority of users. And a dual-A15 at 2.5GHz will beat out a 1GHz PowerPC 460 by a long way.

By the time AmigaOS (or Aros) was fully ported, with the necessary drivers, we would certainly be into A15 systems. A core requirement would be sharing the development of the hardware with other projects (e.g., next generation Pandaboard, Raspberry Pi, etc).

[vidarh]

@Vidarh

None of which makes ARM a great choice for a desktop. The current mobile fad will subside in time. Already things are close to saturated and more and more people are getting tired of the fad all the time. This isnt to say it'll vanish, but there's not much more scope for extra customers. The market is already predominantely made up of trend followers and same people upgrading thier phones, etc.

I find this short sighted at best. Two reasons:

- Get used to a smartphone, and you can't live without it. I use mine to read books, to play games, as my music player, to take notes, as my camera and camcorder, to check bus and train times, for maps, to read my e-mail - sometimes even when I'm at my desk -, to watch videos when I'm traveling, and more. It has consolidated several functions I before used separate devices for (camera, phone, media player) into one and at the same time given me more functionality, cheaper.

This is why the smartphone market is set to overtake the PC market in unit sales *this year*. Last year there were 305 million sold PC's, vs. 296 million sold smartphones. The PC market is stagnant, while the smartphone sales have been continuing to accelerate wildly. This drives massive investment, particularly given that the smartphone market still have vastly higher margins than the PC market, meaning the smartphone manufacturers can already actually afford more R&D than the entire PC industry.

- Higher end smartphones have *already* exceeded the performance, memory and storage capacity of a proven successful market segment for PC's, which means that with HDMI out and a bluetooth keyboard, it is *already* a viable main computer for a segment of users, and this is before Kal El etc. start pushing smart phones and tablets further into this territory.

The *fact* is that the majority of the PC market is now driven by cost conscious users that don't see any point in buying a much faster machine, and so buy machines between $200-$600 and *dropping*. The average price per unit for PC's is around the $400 mark, to the despair of PC manufacturers.

For these users, it does not matter that the CPU's used can't compete with high end x86's, as high end x86's are way out of their price segment, and largely irrelevant to their needs.

The cheapest and most powerful (by a huuuge margin currently,.... despite what people like to claim, arm is a far, far way away from even the most budget of x86 cpus. Even Denver is, when being optimistic, aiming for P4 class performance, something I personally couldnt deal with) will always be best suited to a desktop machine. Currently that's x86.

CPU performance stopped being a big differentiator to the majority of PC buyers at least 4-5 years ago.

[vidarh]

After 3 weeks using a laptop with a 4-core i7 processor and enjoying the speed with which everything happens, I can't see such low power processors having an advantage where a real computer is needed.  Tablets, phones, sure, they need the battery life, but waiting for web pages to finish rendering due to wimpy processors is no fun.

Yet the vast majority of users are unwilling to pay more than ~$500 for a complete computer, so they don't get to see how fast web pages renders on an i7. In the meantime, my 70 GBP (~$110) cheapo Android pad with a 700MHz outdated ARM renders web pages fast enough for casual use for me, and is already about 3 generations beyond the curve.

[takemehomegrandma]

And while ARM doesn't seem to come in a desktop board (that I'm aware of anyway, perhaps I'm just ignorant)

I think a http://pandaboard.org/ would make a perfectly capable desktop machine for most everyday tasks.

I think this is a much better working-out-of-the-box option than the pandaboard, and while perhaps a bit underpowered (about on par with/slightly faster than a Sam440) to be called "desktop", this doesn't stop me and many others from using it as such for everyday usage! And I certainly wouldn't mind this turn into this!

And these devices is soon to be upgraded to the i.MX53 CPU, still Cortex-A8, but much faster and more capable in many ways, and they are doing it *really* cheap (which translates into *less* than $49 for the board, no idea by how much less, but probably considerably less; low cost seems to be the main focus of the development).

When it comes to Cortex-A9 (that we see in various products today, like iPad2, Tegra2 based devices etc), current chips performs on par with the PPC G4 processors, which is not bad at all, given the additional power and benefits of the on-chip support controllers and accelerators that boosts performance of many applications in a smart, resource friendly way. A lot of things will happen in this field soon though, both in clock speed (Tegra3 will be a faster dual core Cortex-A9, have better GPU, etc, and GlobalFoundries will demonstrate a 3GHz dual core Cortex-A9 made with their 28-nm processes sometime in 2012, as well as a low-power 2GHz version), and in number of cores (Freescale recently announced their Quad Core Cortex-A9 based i.MX6, and also showed it off running in real silicon, here it's being benchmarked by Konstantinos Margaritis/Genesi. It may get here sooner than many would expect).

Cortex-A15 brings a whole bunch of new features from the desktop/server world, and a whole new level of performance.

Then we have the partnership between ARM and nVidia, bringing real 64-bit Workstation performance, the "x86 killer", while keeping the ISA backwards compatible. I think others will follow.

ARM will be the only CPU architecture running on *everything* from phones to workstation/servers, with many interesting devices in between.