Quote:

You are most likely correct, this discussion is pointless, unless one of the members here actually points out some valid suggestions as "Pro's" for using ARM instead of x64, and one or more of the MorphOS Dev. Team members happens to read it, and then think . . . . maybe choosing to port to ARM64 isn't such a bad idea. That one team member then talks to the other team members, and they have their own discussion behind closed doors, to make sure they still want to go forward with the port to x64, or if they want to look at possibly porting to ARM64 as well, or instead of, x64. They are free to change their minds, if they have evidence that one choice is better than their current plans.

I still stand by my statement that there are certain "Pro's" to porting to ARM64, instead of x64. I'm not sure if there are enough "Pro's" in favor of ARM64, to outweigh the "Con's", or if the "Pro's" for x64 don't outnumber the "Pro's" for ARM64, but I think it is worth having a discussion about it. Unfortunately, most of the replies to this thread have nothing to do with the discussion I had hoped would take place.

Edit: I don't know how far along the road to porting MorphOS to x64 hardware has progressed, but if it is further than just a concept, and actual x64 code has been written, there is no point in further discussion about the Pro's and Con's of x64, ARM64, or any other possible architecture change. Since MorphOS for x64 is going to be so radically different than our current MorphOS3.11 for PPC, I'm assuming that there is still a long way to go before we will be seeing a public demo of MorphOS for x64, and still longer before a public release and sales begin. I'm also guessing, that any coding work that has begun, can probably be more easily ported from x64 to ARM64, than the job of porting MorphOS3.11 for PPC to MorphOS for x64.

Can we please get back on topic, and discuss the Pro's and Con's of using ARM64 as a future target to port the Next Gen version of MorphOS to, instead of, or in addition to x64 hardware?

[ Edited by amigadave 26.02.2019 - 16:16 ]

There are no pros or cons in the context of a consumer, nor a software developer beyond this:

Is it already a cheap commonplace item in this market segment? yes/no

Stuff like power consumption blah blah is for the interest of hardware developers. In other words, irrelevant. You can have a new CPU arch that actually generates electricity and reverse global warming, but I'm not going to write software for it until it's shipping in all of next year's dells.

> Your argument style is just to scrutinize posts on a line by line basis

Rather on a proposition-by-proposition basis.

> looking for imagined grammatical flaws in casual speech to attack

No, I tend to ignore grammatical and orthographical errors. Not being a native speaker I make enough of them myself after all.

> rather than actually engaging with the meaning of that speech.

What is the meaning of the word "buggy" you used with regard to PPC hardware, if not the presence of bugs?

> you're obsessed with the Efika's ram "not being a bug".

No, I merely stated that this is not an example of the alleged bugs you claimed.

> Nobody said it was

Then why did you mention it as an example in response to my objection of your "buggy hardware" claim?

> it was brought up as an example of how PPC Amigaish systems have generally sucked.

Then you replied to the wrong comment it seems, as my comment you replied to was objecting your "buggy hardware" claim (which you then repeated in comment #21, so apparently you really mean it).

> What does this weird grandstanding over if it's a bug or not get you?

Probably the same that your insistence on "buggy hardware" gets you.

> It's still a shit and useless amount of ram

Yes, I have the very same low opinion of the Efika's RAM amount and stated it numerous times here on MorphZone and elsewhere since its release.

> although now you'll probably [...] dispute the word "useless".

You may have lost the plot here.

> I told you I'm not interested [...]

Indeed you did, but serving your interest isn't my incentive anyway :-)

> you think doubling down on it will get me to take you seriously now.

That's not what I think. Repeating unfounded claims about buggy hardware is doubling down, I think.

> me making fun of you for this

I'm sure you can do better :-)

Quote:

Is this forum site turning into Amiga.org, or AmigaWorld.net, where most of the threads get derailed by useless arguments? (not specifically blaming you Andreas_Wolf, just adding this comment to the end of this thread)

Quote:

Hacking people's posts into singular tiny fragments to interrogate only makes you miss that your point of interrogation is answered in the next few words.

You're ruining your own argument and wasting time saying redundant things because you won't read properly.

[ Edited by khorse 27.02.2019 - 13:27 ]

Quote:

The argument is that that desktop ARM is likely to have the same problems as desktop PPC, namely that those using them are big companies who can design their own support chips and quickly work around any hardware bugs. And will promptly close off their own designs so MorphOS would take years to run on them anyway.

In that context, none of this discussion is off topic.

Here's the thing about PowerPC and MorphOS/AmigaOS: it was tried. And tried. And tried. We aren't relying on any guesses or opinions here.

The original Pegasos fell foul of a northbridge with cache coherency bugs and the fix was expensive and hammered the system's speed. The original AmigaONEs did not fix the issue and much of the board's functionality was replaced by extra cards to avoid triggering them. The hardware had unadvertised bugs.

The Efika was almost useless for desktop because it's SoC had never been designed for it. The hardware was insufficient for intended use.

The Pegasos-2, while faster in every way than the Peg-1, was on paper inferior in several ways because the Marvell just wasn't as advanced as the Articia. The hardware was limited.

The X1000 ran out of the supply of PA6Ts to continue the line, simply because they weren't available any more - and even when they were, they were too expensive.

More recently, the Tabor has been endlessly delayed because every time the board is finalised, some component or another is no longer available.

PowerPC, as a direction for MorphOS, failed because of these reasons: price, availability, power and robustness. It would seem to me that if you want to choose a new CPU family and not fall foul of this again, then you should:

* avoid proprietary hardware
* avoid non-mainstream hardware
* use specifically designed desktop or notebook hardware
* Use only well-tested designs

Right now, that means x64. The only way this will change if desktop ARM flourishes and diversifies... and isn't just controlled by Google.

[ Edited by KennyR 27.02.2019 - 16:08 ]

Completely sound analysis, but prepare to be ignored by the true believer.

>>> it was brought up as an example of how PPC Amigaish systems
>>> have generally sucked.

>> Then you replied to the wrong comment it seems, as my comment
>> you replied to was objecting your "buggy hardware" claim (which
>> you then repeated in comment #21, so apparently you really mean it).

> you miss that your point of interrogation is answered in the next few words.

My point of interrogation was your "buggy hardware" claim referring to any (not some) PPC hardware of the past 20 years, thus including the Apple, bplan/Genesi, ACube and Varisys/A-Eon PPC hardware. I don't see where you answered this. If you did and I missed it, you can surely quote it. (Honestly, your claim is so ridiculous that I cannot even imagine how it could be proved seeing as it could be falsified by no more than one single counterexample.)
My opinion on this: The PPC hardware designed by professional hardware designers like Apple, Varisys or bplan is not significantly more buggy than any random x86 hardware.

> You're [...] saying redundant things

Yes, I know that I'm somehow repeating myself in my replies to repeatedly uttered nonsense claims.

> you won't read properly

Apparently it's you who didn't read properly what he replied to as you failed to mention any bugs of Apple, bplan/Genesi, ACube and Varisys/A-Eon PPC hardware in comment #15. The only part of your comment that is contentually connected to what you replied to is your mention of the Peg1's "bad chipset", the bugs of which I still believe are remedied by the April2 in connection with FSB clock reduction. And even if they are not, the Peg1 is just one single of a three-digit amount of (desktop/mobile) PPC hardware released in the past 20 years.

> The original Pegasos fell foul of a northbridge with cache coherency bugs
> and the fix was expensive and hammered the system's speed.

Fully correct, the fix of the bugs came at a cost.

> The original AmigaONEs did not fix the issue and much of the board's
> functionality was replaced by extra cards to avoid triggering them.
> The hardware had unadvertised bugs.

Exactly. In contrast to almost all other PPC hardware of the past 20 years, the Eyetech AmigaOnes were overly buggy.

> The Efika was almost useless for desktop because it's SoC had
> never been designed for it.

The whole board had never been designed for it. In my opinion, even more restricting than the low CPU performance was the laughable amount of RAM, although the SoC could handle at least twice the amount (which would still have been poor).

> The hardware was insufficient for intended use.

Intended use according to Genesi: "embedded, thin client, industrial applications [...]. Example product solutions include: Thin-Client for X or Windows Terminal Server (rdesktop) displays; Home Theater satellite, to relay recorded or downloaded TV shows and movies around the home; High power-use/performance-ratio cluster node; Industrial control and robotics; VOIP/Video Phone connected to a TV; Webcam Security solution".

> the Marvell just wasn't as advanced as the Articia.

The Marvell had a GbE controller and could handle DDR RAM while the Articia was stuck at SDR RAM :-)

> The hardware was limited.

Yes, biggest problem in my opinion was that it took the Peg2 until the last board revision to remove the 1 GiB RAM limit.

> every time the board is finalised, some component or another is no longer available.

The Tabor had been finalized for years. It was only when OS4 for the board (not the board itself) was recently finalized by Hyperion, A-Eon realized that some (unspecified) component(s) was/were no longer available in the quantities required for production run so had to be replaced, which in turn required a board redesign to unspecified extent.

> PowerPC, as a direction for MorphOS, failed because of these reasons:
> price, availability, power and robustness.

Can you elaborate on the "robustness" part of that statement?

> Completely sound analysis

Biggest part of the "analysis" I agree with, just some minor objections (like intended use of Efika 5200B and Marvell vs. Articia advancement) and a request for elaboration ("robustness").

> prepare to be ignored by the true believer.

He isn't being ignored by me at least :-)

> desktop ARM is likely to have the same problems as desktop PPC, namely
> that those using them are big companies who can design their own support
> chips and quickly work around any hardware bugs. And will promptly close
> off their own designs so MorphOS would take years to run on them anyway.

I'm not sure that "desktop ARM", in contrast to old "desktop PPC" (aka G3/G4/G5), will necessarily have "their own support chips" given the trend towards more and more integration. As a result, "desktop ARM" could simply lack those means to lock out any not their own OS as the I/O controllers etc. are part of the "desktop ARM" SoC documented by its vendor for 3rd parties.
Also, I'm not sure that, in case of "desktop PPC", "their own support chips" were a real obstacle for porting any not their own OS to it ;-)

Quote:

IIRC it was advertized to be available in configurations up to 1 GiB RAM. Dunno, if such a configuration was ever tested - and I rather doubt it (would have needed changes in smart firmware).
With 256 MiB the board would have been way more usable. But it's all void and grey theory now, the Efika is EOL and since Apple kit support there aren't many that cry actual tears about it anymore.
I still like my Efika though.

>> the SoC could handle at least twice the amount

> IIRC it was advertized to be available in configurations up to 1 GiB RAM.

This would have been dishonest advertising then, considering this:

"256-MByte addressing range per Chip Select (Two CS lines available)"
https://www.nxp.com/docs/en/reference-manual/MPC5200BUM.pdf (p. 15)

"256 MB addressing range per CS, two CS available"
https://www.nxp.com/docs/en/data-sheet/MPC5200BDS.pdf (p. 1)

"256 Mbyte addressing range"
https://www.nxp.com/docs/en/product-brief/MPC5200BPB.pdf (p. 3)

So the maximum could have been either 256 MiB or 512 MiB.

Quote:

Maybe my memory is mixing this up with the Efika512x announcements. Which, eventally doesn't matter the slightest, as Efika5200B is EOL and Efika512x never did it (and the show stopper "cache coherency" wouldn't have been too much of an issue for MorphOS AFAIK)... Pity as it held really some potential back in ints time. But wel, long ago and my stance is not to look back too much, but rather forward! Thing is ppc is in limbo/coma and IMHO not a path worth to pursue any further. The Raptor boards could be worth a _little_ thinking though, but *I* would put all efforts into MorphOS NG on x64. Current MorphOS is just pretty polished and should still get some love for the next couple of years (i.e. some updates of this and that), but focus should be the inevitable break and switch to x64.
It's shitload of work but eventually one thing is cystal clear: On PPC MorphOS will not have a future, on x64 there's at least a little prospect for some sustainability.

Quote:
Yes, that's the point.

I also do 'like' ARM/ARM64 more than amd64, but I don't expect any other person to feel just the same or write some code for it.

Quote:

It doesn't matter what I write because your response always looks like this:

Quote:

I have actually already written something that, if you'd had bothered to read, would mean you didn't need to ask this. But because you don't actually read anything, you haven't noticed.

Like, in my last post I already I proved you don't read what you reply to before you reply to them. Otherwise you wouldn't question posts that answer your question in the next sentence.

Hint: Read posts all the way from the beginning to the end, only reply after you finish doing that. Otherwise you're just interrupting before you understand, guessing at it's meaning and getting it wrong, and asking questions of it that are answered by it.

[ Edited by khorse 01.03.2019 - 18:14 ]

>>> you miss that your point of interrogation is answered in the next few words.

>> My point of interrogation was your "buggy hardware" claim referring to any (not
>> some) PPC hardware of the past 20 years, thus including the Apple, bplan/Genesi,
>> ACube and Varisys/A-Eon PPC hardware. I don't see where you answered this.
>> If you did and I missed it, you can surely quote it. (Honestly, your claim is so
>> ridiculous that I cannot even imagine how it could be proved seeing as it could
>> be falsified by no more than one single counterexample.)

> It doesn't matter what I write

As long as you refuse to back up your ridiculous claim, it really doesn't matter what you write in your attempt to further distract from it. This has been true since your very first reply to me in this thread. It was never about anything else than your ridiculous claim in my replies to you.

> There is no interrogation because we are not at a police station.

You introduced this word to this discussion. I should better have put it in quotation marks to reflect that I merely adopted it.

>> I don't see where you answered this.

> What don't you?

I don't see where you answered my point of "interrogation".

>>> If you did and I missed it, you can surely quote it.

>> Quote what?

Your answer to my objection of your "buggy hardware" claim.

> I have actually already written something that, if you'd had bothered
> to read, would mean you didn't need to ask this. But because you
> don't actually read anything, you haven't noticed.

I just re-read everything you wrote in this thread and I surely can find "something" (whatever thing that is) but I can't find your supporting arguments for your "buggy hardware" claim. Can you please be so kind and point me to them?

> in my last post I already I proved you don't read what you reply
> to before you reply to them.

I don't think this is what you did (proved anything, that is). What I sincerely believe is that you didn't properly read what you replied to in your first reply to me and are now too ashamed to admit it, thus trying your best to distract from the original point of disagreement (i.e. your "buggy hardware" claim and my objection to it) and to turn this into a nonsense discussion about me instead.

> Otherwise you wouldn't question posts that answer your question in the next sentence.

Please point me to that "next sentence" that answers my objection of your "buggy hardware" claim.

> Read posts all the way from the beginning to the end, only reply after you finish doing that.

That's exactly what I do. Thanks for approving my approach.

> you're just [...] guessing at it's meaning and getting it wrong,
> and asking questions of it that are answered by it.

I repeat my question from comment #28: What is the meaning of the word "buggy" you used with regard to PPC hardware, if not the presence of bugs?
My objection to your ridiculous "buggy hardware" claim surely isn't answered by the claim itself. I'm sure you can see the logical fallacy of this.

Quote:

This has nothing to do with this thread, so please take your arguments about how some member replies to your posts to private mail, and out of this forum thread. Same goes for you Andreas_Wolf, replies to this line of arguments does not add to this discussion.

Can a moderator please step in and stop this nonsense?

[ Edited by amigadave 01.03.2019 - 12:00 ]

> Maybe my memory is mixing this up with the Efika512x announcements.

Could be. I don't remember if the RAM amount was part of the announcement, but yes, the MPC5121e can address at least 1 GiB RAM.

> the show stopper "cache coherency" wouldn't have been too much of an issue
> for MorphOS AFAIK

Support for cache-incoherent SoCs was added with MorphOS 3.8 for the PPC460EX, so I think that yes, this support could as well have been implemented years before for the MPC5121e.

Quote:

You don't need to worry, in his last post he didn't notice when I did an impersonation of him, even though it was in a quote block and directly demonstrated a behavior of his that I'd described in depth three times. He just replied to it like it was my own voice.

It's only then I realized he's got a disability or sickness or something, it'd be unfair of me to pick on him any further.

It is time to stop this old crap "ppc is dead".

This whole discussion is simply stupid.

It is obvious that after breaking binary compatibility with 68k future Morphos should be made hardware independent.

Make Morphos open source on LGPL and let everyone use Morphos on whatever hardware they want.

> I [...] demonstrated a behavior of his

Only that you did it wrong. You can PM me for instructions on how to do it right.

> He just replied to it like it was my own voice.

And I could easily answer every single point of it, whereas you keep refusing to answer mine.

> he's got a disability or sickness or something

I think what you've been exhibiting in this thread is called intellectual dishonesty or something.

> it'd be unfair of me to pick on him any further.

Don't worry, I don't feel picked on me by you. I just feel you're making a complete fool of yourself by refusing to back up your ridiculous "buggy hardware" claim and trying your best to distract from the original point of disagreement (i.e. your "buggy hardware" claim and my objection to it) and to turn this into a nonsense discussion about me instead.

Quote:

I don't think there'd be any point in going multiplatform and breaking 68k compatibility. None of the features that could be gained would outweigh the damage of losing the software library.

Besides which, an intel port would mean multiplatform by definition. It wouldn't make sense to throw away the existing work and alienate the entire user base.

The problem isn't so much the processor architecture, it's the platform.

You need hardware that's available, that you can get up and running on, and will be around for a while.

For desktop and laptops thats mostly x86.
There are some Arm laptops, they're not as powerful as PC laptops yet but they are improving rapidly. The first gen Windows 10 laptops used tablet processors but the next get will have proper laptop processors.

Arm has now announced a high end processor (the snappily titled Neoverse N1) but that's really for servers so I don't expect you'll see them in laptops, but who knows.

However, computing isn't just about desktops or laptops these days. There's computers of one sort of another in all sorts of form factors for all sorts of prices.

There's lots of little machines around these days like the Android TV boxes which are probably rather more powerful than your G5s by now.

Then there's the Raspberry Pi, they've sold 25 MILLION of those things! I think they'll be around for a while and they're really cheap.


There's platforms available for x86 and Arm so it really comes down to what is desired, do you just want new faster machines for the existing user base or do you want to grow the user base. I expect Raspberry Pi would be a better choice for growing the user base as you'l have lots of young people happy to experiment.