MiniMig + PowerPC = OS4 Hardware for all

[Hans_]

downix wrote:
I say, run the classic AmigaOS in a sandbox, much like OS X runs the classic MacOS, and build a new platform.  But, the time to market scares away everyone, and nobody seems willing to take the risk.

That's exactly what Hyperion want to do (or should I say, the OS 4 dev team). Even if they use a similar API, based on the same principles, they have to break compatibility in order to implement things like memory protection and multiprocessing.

Hans

[downix]

Hans_ wrote:

downix wrote:
I say, run the classic AmigaOS in a sandbox, much like OS X runs the classic MacOS, and build a new platform.  But, the time to market scares away everyone, and nobody seems willing to take the risk.

That's exactly what Hyperion want to do (or should I say, the OS 4 dev team). Even if they use a similar API, based on the same principles, they have to break compatibility in order to implement things like memory protection and multiprocessing.

Hans

Indeed.  My main worry is in hardware, namely the complete lack of forward thinking.  You either have purists that insist on sticking to Mot (PPC or Coldfire or m68k), or genericists that want to go x86 (equally a dead-end in my view).  I truely want the community to tell these corporate morons to shove off and carve our own path, independent of corporate overlords dictating the direction we can go in.  That's why I'm heavily for licenseable CPU architectures.

Hard to believe we're on the same side now, eh?  

[koaftder]

freqmax wrote:
@downix:
The reason for Minimig is that the hardware is dying physicaly. And there is no new ones made.
Software emulation won't give cycle accurency that demos & games need.

As for CPUs, ARM is trigger happy litigator. I would be careful about being dependent on them.

Hardware doesn't provide an advantage when it comes to cycle exact emulation, all hardware can do is provide a speedup. If you can't model something accurately in software, hardware certainly can't come to the rescue as the model has to come first.

[Skyraker]

THIS the A2060 we're talking about?


*edit*.. no it's not... got this from BBOAH though.


CE: Blizzard 2604
Connects To: CPU Fast Slot (A2000)

DUAL Processor:   060@50Mhz and PPC604e@180Mhz or 200Mhz
FPU:   Internal for both processors
MMU:   Internal for both processors
Max Ram:   128MB
Ram Type:   4 x 72pin SIMM sockets.

An accelerator which was designed to plug into the CPU Fast Slot of the A2000. Unfortunately this card never got past the prototype stage and so it was never released to the public. Since Phase 5 went bust, DCE bought the rights to all their designs but it is unlikely that they will ever release this card. There exists PowerPC cards for other models of Amigas such as the Cyberstorm 604e PPC and the Blizzard 603e PPC. The idea is that the PowerPC can be used to run software which supports it, much faster than any 68k processor can, with the 68k providing compatibility with existing 68k Amiga software. There is a surprisingly large amount of PowerPC software for the Amiga, and even alternative PowerPC operating systems such as Linux which would run on this card (had it been released). The card also includes a SCSI-3 Ultrawide (68pin) controller and an expansion port for the Cybervision PPC graphics card.

This card wasn't released by Phase 5, even though they did design it. It may be released by DCE.

[Kronos]

@HenyCase

Don't know, what it is with you and the PPC-developer-board, but there were probraly only a dozen or so ever made, and I know of only 2 in the hands of community memebers.

The Blizzard2060 is also in short supply ( :cry: why oh why did I sell mine  :cry: ), so thats a no go.

@Skyraker

Bringing the Blizzard2604 to market (if it was ever finished that is) is just as "easy" as bringing back the CS-PPC and BPPC ..... you know, non-ROHS compatible out-of-production components to be mounted in very expensive production-methods
 :roll:

[little]

That's why I'm heavily for licenseable CPU architectures.

That sounds expensive :S One thing is certain, we need to focus in what can be achieved versus what might be cool/right/visionary but is not feasible in the near future to be done by the community. Yeah I know I should follow my own advice ^^;

[downix]

little wrote:

That's why I'm heavily for licenseable CPU architectures.

That sounds expensive :S One thing is certain, we need to focus in what can be achieved versus what might be cool/right/visionary but is not feasible in the near future to be done by the community. Yeah I know I should follow my own advice ^^;

http://www.sparc.org  $100 licensing fee.

http://www.srisc.com/?s1 64-bit performance SPARC core design, GPL'd, free.

Doesn't seem much more expensive than the FPGA MiniMig to start with.

[little]

Doesn't seem much more expensive than the FPGA MiniMig to start with.

Oh, but there is "more than meets the eye". Using a non-standard 64 bit core in the minimig FPGA brings any immediate benefits?
------------------------------------------------
Lets break it down what sparc has to offer.

Pros

- Good CPU architecture
- Not compatible with 68k

Cons

- Non standard. Most programmers are not interested in learning another CPU architecture (most are familiar with one only).
- No simple retail option to implement a non-FPGA version.
- Low speeds (unless you get a very expensive FPGA or mass produce it, in which case it becomes expensive).
------------------------------------------------
Not lets see the freescale 68k

Pros

- Well known architecture, many programmers and hardware engineers are familiar with it. Software cores are highly feasible.
- Compatible with motorola 68k line.

Cons

- Expensive hardware. Only the 68000 is cheap, price increases quite a bit for later versions. I am speaking about retail prices, if somebody can get it used at a lower price or even free does not guarantee everybody can get those prices.
- Low speeds
------------------------------------------------
Now lets see coldfire

Pros

- Easy to implement architecture. Partially compatible with 68k, just better.
- Cheap
- Embedded. Some versions include usb, ethernet & pci right on the chip.

Cons.

- Low speeds. Higher than 68k, on par with geode but still less than current procesors.
------------------------------------------------
Now lets see PPC

Pros

- Well known architecture
- Cheap
- Embedded.

Cons

- Not compatible with 68k
- Low speeds. better than coldfire,but still far from current speeds.
------------------------------------------------
Now lets see x86

Pros

- Well known architecture
- Fast
- Cheap

Cons

- Not comatible with 68k
- Needs lots of external support chips.

[downix]

little wrote:

Doesn't seem much more expensive than the FPGA MiniMig to start with.

Oh, but there is "more than meets the eye". Using a non-standard 64 bit core in the minimig FPGA brings any immediate benefits?
------------------------------------------------
Lets break it down what sparc has to offer.

Pros

- Good CPU architecture
- Not compatible with 68k

Let's add one, guaranteed source through licensing.  If you license the arch, no "cutting off at the knees" by vendor lockout[/quote]

Cons

- Non standard. Most programmers are not interested in learning another CPU architecture (most are familiar with one only).
- No simple retail option to implement a non-FPGA version.
- Low speeds (unless you get a very expensive FPGA or mass produce it, in which case it becomes expensive).
[/quote]mass production is not as expensive as it once was.

------------------------------------------------
Not lets see the freescale 68k

Pros

- Well known architecture, many programmers and hardware engineers are familiar with it. Software cores are highly feasible.
- Compatible with motorola 68k line.

Cons

- Expensive hardware. Only the 68000 is cheap, price increases quite a bit for later versions. I am speaking about retail prices, if somebody can get it used at a lower price or even free does not guarantee everybody can get those prices.
- Low speeds
------------------------------------------------
Now lets see coldfire

Pros

- Easy to implement architecture. Partially compatible with 68k, just better.
- Cheap
- Embedded. Some versions include usb, ethernet & pci right on the chip.

Cons.

- Low speeds. Higher than 68k, on par with geode but still less than current procesors.
------------------------------------------------
Now lets see PPC

Pros

- Well known architecture
- Cheap
- Embedded.

Cons

- Not compatible with 68k
- Low speeds. better than coldfire,but still far from current speeds.
------------------------------------------------
Now lets see x86

Pros

- Well known architecture
- Fast
- Cheap

Cons

- Not comatible with 68k
- Needs lots of external support chips.

And the issue with all of the bottom ones is, vendor lockout.  Motorola unable to supply the chip?  You're up the creek.  Can't find a northbridge, up the creek because you can't license the CPU bus in order to have one made.  Intel doesn't supply you with docs fast enough, you are locked into older generations while your competition surges forward.

No solution is a perfect one, but I would sooner dedicate myself to being able to stand on our own two feet than being locked into some corporate overlords whim.

[HenryCase]

Skyraker wrote:
An accelerator which was designed to plug into the CPU Fast Slot of the A2000. Unfortunately this card never got past the prototype stage and so it was never released to the public. Since Phase 5 went bust, DCE bought the rights to all their designs but it is unlikely that they will ever release this card.

Thank you for your post on the Blizzard2060 Skyraker, very informative. I didn't realise DCE bought the rights to the designs, I thought bPlan may own it.

Going to investigate further now.

[AJCopland]

There's precedent for trying out other architectures too. We've always been able to run hardware emulators in amigas since the early days with PC286 support. Does anyone remember the Alpha 21164 CPU which the Siamese team were trying to build a machine around? :-D

First and foremost the MiniMig v1.x is an A500/A600 emulation machine. It doesn't have to stay that way and ideas like this are really great to see even if they are hard to do.

Andy

[HenryCase]

Kronos wrote:
@HenyCase

Don't know, what it is with you and the PPC-developer-board, but there were probraly only a dozen or so ever made, and I know of only 2 in the hands of community memebers.

The Blizzard2060 is also in short supply ( :cry: why oh why did I sell mine  :cry: ), so thats a no go.

@Skyraker

Bringing the Blizzard2604 to market (if it was ever finished that is) is just as "easy" as bringing back the CS-PPC and BPPC ..... you know, non-ROHS compatible out-of-production components to be mounted in very expensive production-methods
 :roll:

Kronos, please let me explain my plan, as this will hopefully help you see why the Blizzard 2060 might not be the dead end you describe it as.

Clearly the hardware as it stands now is not suitable, for the following reasons:
1. As you state, there aren't enough of them around.
2. The hardware, whilst almost fully developed, doesn't exist in a finished state.

To get around these issues, I have been trying to contact the patent owners to see if we can come to an agreement about releasing the technology to the Amiga community (i.e. open sourcing the h/w and software in its existing state). Considering how close to complete the h/w is, and the fact that it runs OS 3.9 pretty well I think it would be a great way to have an affordable OS4 machine in the hands of the community.

Open source the hardware you say, are you mad? Well, if you don't ask, you don't get. Certainly the current h/w owners don't have a viable use for the tech, I am hopeful I can come to some agreement. I e-mailed bPlan thinking that they may hold the payments (got a reply too, still early days mind), but I'll contact DCE too after Skyraker told me they hold the rights. Not promising anything, which is why I kept this secret before.

Hope that explains my madness!  :-D

[little]

downix wrote:
Let's add one (feature about sparc), guaranteed source through licensing.  If you license the arch, no "cutting off at the knees" by vendor lockout

Now that you mention it, Coldfire CPUs have the same advantage since 2006

mass production is not as expensive as it once was.

Yet mass producing the minimig and indepently mass producing a CPU would duplicate costs, which for a project such as this at this point in time (and the near future) is a big no-no.

[/quote]And the issue with all of the bottom ones is, vendor lockout.  Motorola unable to supply the chip?  You're up the creek.[/quote]
I think CPUs produced by freescale (except the ARM) are open to licensing, so I think what you really tried to say "Intel"

Can't find a northbridge, up the creek

Let's then not forget some Coldfire and PPC CPUs have the northbridge and southbridge built into it.

Intel doesn't supply you with docs fast enough, you are locked into older generations while your competition surges forward.

I do not think the minimig (be it v2, v3, etc.) will need to get the fastest intel cpus, as a matter of fact it is quite unlikely since Intel cpus are good for software emulation due to the high speeds they attain, but I do not think somebody has seriously thought they can be in the future part of the minimig.

No solution is a perfect one, but I would sooner dedicate myself to being able to stand on our own two feet than being locked into some corporate overlords whim.

ATM the only CPU overlord is Intel, AMD is heavily wounded (I believe apple will buy them when the time is right), Motorola left the scene and created Freescale to make some a little profit and everybody else has hardly any market share.

[Hans_]

downix wrote:
Indeed.  My main worry is in hardware, namely the complete lack of forward thinking.  You either have purists that insist on sticking to Mot (PPC or Coldfire or m68k), or genericists that want to go x86 (equally a dead-end in my view).  I truely want the community to tell these corporate morons to shove off and carve our own path, independent of corporate overlords dictating the direction we can go in.  That's why I'm heavily for licenseable CPU architectures.

AFAIK, PowerPC is licenseable as well. IBM and Freescale own it (I think that IBM may have more say), AMCC and PASemi are licensed to use it, as is Xilinx, for their Virtex-4 FPGAs. I have no idea what the licensing costs are.

In the end, the actual CPU won't matter too much. The old OS will be sandboxed either way. So long as we make the jump to 64-bit, it's all good. However, a 64-bit PowerPC CPU would make the transition easier as there's no emulator to write. Forget buying PowerPC chips from Freescale; there are other vendors with better implementations.

Hans

[little]

So long as we make the jump to 64-bit, it's all good.

Slowdown, that is far away into the future.

However, a 64-bit PowerPC CPU would make the transition easier as there's no emulator to write.

If you say so because AmigaOS 4 is to be released any minute now, think again. Hyperion cannot do any upgrades to the operative system and they are in litigation with Ainc and whoever becomes the winner might not support a community created amiga platform. I think we are better of with AROS, which is open source (like the minimig) and therefore there are no litigation hassles/company CEOs to worry about.

Forget buying PowerPC chips from Freescale; there are other vendors with better implementations.

The keyword is not better, the keyword is cheaper

[downix]

Hans_ wrote:

AFAIK, PowerPC is licenseable as well. IBM and Freescale own it (I think that IBM may have more say), AMCC and PASemi are licensed to use it, as is Xilinx, for their Virtex-4 FPGAs. I have no idea what the licensing costs are.

In the end, the actual CPU won't matter too much. The old OS will be sandboxed either way. So long as we make the jump to 64-bit, it's all good. However, a 64-bit PowerPC CPU would make the transition easier as there's no emulator to write. Forget buying PowerPC chips from Freescale; there are other vendors with better implementations.

Hans

And still no northbridges nor ones with integrated modern northbridges.  I looked into licensing the PowerPC, with no luck.  I also looked at Coldfire licensing, no go.  You can make a clone, as some folk have done, but if you're doing that, might as well clone the 68k directly.

At this time, SPARC remains the best choice, in my opinion, for a next-generation platform.