@Yasu

With such small market, price is always an issue - be it PowerPC motherboard or FPGA board with 1/20 of its performance. Knowing A-Eon ordered production of 1000 Tabor motherboards for AmigaOne A1222, we will probably see lowest price level achievable in such conditions.

@pavlor

I have no problems with any product from a hobbyist perspective.

_________________
Amiga Forum - Sweden's best Amiga Magazine

My MorphOS Blog

"Free speech includes other peoples right to offend you."

@itix

Quote:

itix wrote: @cdimauro Quote: In both cases (only compilation needed, and coders support to the SPE), there's the ABI question which itix already raised, and should be solved. I gave a solution before, but it slows down the execution. There is one more complex case... Consider you have following library code compiled for the FPU: double foobar(double a, double b) { return a + b; } Next you are compiling your binary for SPEFP and want to use this foobar() function from the library (shared, static, or dynamic): void main() { // Compiled using SPE floating points double x = 1.0, y = 2.0; // Call our FPU floating point function with SPEFP params double rc = foobar(x, y); // Ooops? } Do you see the problem?

I think so (it should be the same scenario which I was talking about in some previous comment), and that's a mess.

The compiler should distinguish between the calling / SPE code and the library / FPU one, and generate ad hoc instructions to transfer the SPE registers contents to the corresponding FPU "registers". This slows down the execution (because a normal SPE store + a trapped FPU load should be execute).

But at least it works, albeit running then the API code is quite slow, since the library uses the trapping mechanism for the FPU.

Now imagine that you want to speed-up it, having also the library compiled with the SPE unit support(so, no trapping involved): it's nightmare.
Quote:

Quote: - "string" instructions are also unsupported and trap every time; That is not big deal. In GCC it is not automatically enabled and to use those instructions you must pass additional flag to the compiler. It is just matter grep'ng makefiles and not using string instructions anymore.

I wasn't referring to it, but to existing applications that already used such instructions. Here the trap mechanism is used for them.

Of course, if you're able to recompile the application, you have no problem, since they aren't generated.


@OlafS25

Quote:

OlafS25 wrote: @cdimauro also posted on another thread already... that is from annual report of Applied Micro Circuits. It explains why new hardware from a different supplier developed. And it shows the decline of PPC in general. [...] Source: http://www.sec.gov/Archives/edgar/data/711065/000144530514002375/amcc331201410-k.htm

Olaf, that's something that I repeat from several years, and I was blamed and persecuted for that, in some fanatic forums.

Now there's a public, official statement, but I doubt that it'll change something: people continues to imagine a future for PowerPCs, because... they have new machines with such processors.
Quote:

BTW some confusion... is Sam 460 now dropped or not?

AMCC will not drop her PowerPCs line (for several years), so this isn't the case.

But the path is clearly traced...


@AmigaBlitter

Quote:

AmigaBlitter wrote: @OlafS25 https://www.power.org/newsletter/power-org-september-2015-newsletter/

Did you see any new PowerPC microarchitecture here?

BTW, the numbers that they give are quite obsolete. Take a look at the Xeon processors that went out in the last months.


@Yasu

Quote:

Yasu wrote: From what I understand (which isn't saying much) x86/AMD64 is fast but power hungry,

Essentially it depends on the transistor budget for the decoder, since this is the most complex and one of the most active unit in the x86/x64 processors.

However pay attention that the decoder doesn't exponentially scale with the number of transistors packed on the chip. In fact, the transistor budget isn't increased so much from long time, and it's around some millions transistor for it.

So, its impact was significant when the chips were made of millions and tenths of millions of transistor, but with modern chips which pack billions of transistor they haven't anymore a sensible impact.

Last but not least, such CISCs have more complex instructions which make more "usuful" work compared to the simpler RISC instructions. This helps both execution speed (e.g.: more operations performed) and code size (much dense code). Code density is also another important factor, which impacts in all memory hierarchy, from the L1 (code) cache down to the system memory, and this also impacts to the power consumption.
Quote:

ARM is a lot slower but very power efficient and PPC is suppose to be very fast but reasonably power efficient. Problem is that all processors are getting faster and faster, even ARM, making PPC pretty much obsolete as it simply can't keep up.

Exactly. That's because they are made of A LOT of transistor. And that's why the so called "x86 tax" is quite less significant from long time (see above).

@Yasu

Quote:

Basically, PPC is screwed and we can maybe even consider ourself lucky that we can buy any types of processors today.

I think I may have lost the plot here (which is easy to do in a 30-season soap opera), but isn't Hyperion stuck with PPC as a platform because they don't have a license to port AmigaOS to x86_64 or ARM? If they don't have a license for a port, is there something preventing them from getting one? If they can get a license, do they get it from Amiga Inc. or Cloanto?

If this can't be untangled (preferably before I die) it seems like the future may be PPC emulation on x86_64.

Update: I should not fail to mention AROS as a way forward as well, since that already runs directly on x86_64, at least on some hardware.

Last edited by bison on 20-Oct-2015 at 06:56 PM.

_________________
"Unix is supposed to fix that." -- Jay Miner

@bison

is still the current theory

Of course the Acer theory is still out there too, but the notion that alice is an Acer would not support that theory in any way. Chalk that up to coincidence.

#6

Last edited by number6 on 20-Oct-2015 at 06:46 PM.
Last edited by number6 on 20-Oct-2015 at 06:45 PM.

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@number6
Thanks for the link. It seems like the situation is even more complicated (and obscure) than I thought.

At this point I'm sure someone has asked Hyperion point blank if they are legally entitled to port AmigaOS to x86_64. My question is: did they answer directly ("yes" or "no"), or did they equivocate? If the answer was "no", did they give a reason?

_________________
"Unix is supposed to fix that." -- Jay Miner

@bison

Quote:

think I may have lost the plot here (which is easy to do in a 30-season soap opera), but isn't Hyperion stuck with PPC as a platform because they don't have a license to port AmigaOS to x86_64 or ARM?

The settlement agreement between AInc and Hyperion does not mention PPC, it doesn't specify any CPUs or hardware platforms at all and puts absolutely no limits on what Hyperion can do with OS4.

There are people behind the scenes who keep speculating about the "spirit of the contract", implying that a lawyer could somehow argue since OS4 was PPC only at the time, the contract not explicitly stating the possibility of a port means OS4 needs to stay PPC only. Personally, I don't see it - but I stopped assuming that "they can't be that dumb" nearly decade ago.

There's a more likely scenario though, as I explained in the posting #6 so kindly linked to:

Hyperion has also licensed the code that was written for OS4. Most if not all of the OS4 developers still own their code, Hyperion just has a license to distribute it. We don't know what conditions these contracts contain - some or all of that code might only be licensed for PPC.

Again, prime suspects for something like that would be the Friedens. Since they're allegedly owed substantial amounts of money, that would be very bad news for anybody hoping for an x86 port.

Quote:

If they can get a license, do they get it from Amiga Inc. or Cloanto?

Amiga Inc. can not hand out licenses to any Amiga code anymore, they even had to license Kickstarts from Cloanto themselves.

Last edited by cgutjahr on 20-Oct-2015 at 07:05 PM.

@number6

Quote:

Of course the Acer theory is still out there too

Somebody had actual lawyers look into that, AFAIK. They came back empty.

Time to put your tinfoil hat back on

@cdimauro

Thanks for the info

@bison

One AmigaOS developer (can't remember who) said that the kernel is hard coded to work with big endian. Short of a complete rewrite of the kernel (and other code?) a port to a little endian system isn't apparently possible. Sure, there are bi-endian ARM processors, but from what I've read they are on the way out. So choosing any of them isn't safe either. That leaves little endian processors and also the basic problem untouched.

_________________
Amiga Forum - Sweden's best Amiga Magazine

My MorphOS Blog

"Free speech includes other peoples right to offend you."

@cgutjahr

Oldest public record on AW I can find atm regarding public statement:

Source

I should add this as well:

It may surprise you (and others) but I would really like to port AmigaOS 4.1 to x86.-HJF

However in 2006:

Reason 1: Hyperion does not have a licence for x86, only for PowerPC.

#6

Last edited by number6 on 20-Oct-2015 at 07:32 PM.
Last edited by number6 on 20-Oct-2015 at 07:28 PM.

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@Yasu

Quote:

One AmigaOS developer (can't remember who) said that the kernel is hard coded to work with big endian. Short of a complete rewrite of the kernel (and other code?) a port to a little endian system isn't apparently possible.

I suppose exec would have to be rewritten yet again.

_________________
"Unix is supposed to fix that." -- Jay Miner

@cgutjahr

Quote:

Hyperion has also licensed the code that was written for OS4. Most if not all of the OS4 developers still own their code, Hyperion just has a license to distribute it. We don't know what conditions these contracts contain - some or all of that code might only be licensed for PPC. Again, prime suspects for something like that would be the Friedens. Since they're allegedly owed substantial amounts of money, that would be very bad news for anybody hoping for an x86 port.

This I did not know. It seems like an intractable situation with not much chance of a good long-term outcome.

Anyway, sorry to wander so far off topic.

_________________
"Unix is supposed to fix that." -- Jay Miner

@number6

I don't get it. How can they not have a licence? Wasn't AmigaOS 4 granted an exclusive license with the 2009 court settlement with Amiga Inc.? Sure such an agreement isn't still valid?

_________________
Amiga Forum - Sweden's best Amiga Magazine

My MorphOS Blog

"Free speech includes other peoples right to offend you."

@Yasu

My fault for not putting every link in by order of date.

The last link is the oldest reference, prior to lawsuit and settlement.(2006)

#6

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@number6

Quote:

number6 wrote: However in 2006: Reason 1: Hyperion does not have a licence for x86, only for PowerPC. #6

Reason 3 and 4 don't apply anymore.

@number6

So ... it's still valid?

_________________
Amiga Forum - Sweden's best Amiga Magazine

My MorphOS Blog

"Free speech includes other peoples right to offend you."

@Yasu

Anything stated before 2009 Settlement about Hyperion´s rights to OS4 is probably not valid anymore - especialy if Settlement Agreement states otherwise. You may read said document yourself.

@cgutjahr

Quote:

Time to put your tinfoil hat back on

Time to put this type of statement to rest.
Its meant to be insulting.
And further, it appears to confuse a paranoid state of mind with schizophrenia.

Just because your theory appears paranoid, it doesn't make you wrong OR crazy.

Although, personally, as a biased person I would prefer to believe anything negative I hear about Hyperion, I KNOW most are not true.

Lets face it, Hyperion has NOT made a fortune on OS4 (at this point I'd be surprised to hear that it has generated a profit), SO maybe they are just fanatics like the rest of us.

AND, if Cyborg is to be believed, even Hyperion's developers are not completely pleased with the choice of the P1022 for Tabor.

Last edited by iggy on 20-Oct-2015 at 08:33 PM.

@iggy: who decided to take the P1022? I don't think Trevor, which seems to have no technical background.

Some technicians made it, and they should know the limits and problematics of choosing an e500v2.

And you don't select a processor from the day to night: you have time to think about it, and to the consequences.

As usual, Trevor had bad counselors. Maybe the next time (if there'll be another AmigaOne project) he has to find better ones...

@cdimauro

Quote:

he has to find better ones...

You?