Quote:

I didn't realize how "out of touch" you were with the MorphOS community, and the several discussions that have taken place regarding the announcements several of the MorphOS Dev. Team members have made about the decision to break away from AmigaOS 68k compatibility (except perhaps through the inclusion of EUAE in some form), as well as moving on to 64bit on the amdX64 architecture. The decision has already been made, and probably some work in that direction has already been completed. This discussion is for people who want to point out the Pros and Cons of choosing to support the ARM64 architecture (or others), instead of, or in addition to the already made choice of x64.

Quote:

Finally a useful reply!

I had been thinking "either or", for the most part, instead of making the next generation of MorphOS cross platform with a simple recompile. Different drivers would still be needed, but I'm pretty sure that the MorphOS Dev. Team will create an OS that is much more flexible than the constraints we have been tied down with, by starting out with AmigaOS3.1 API compatibility.

Your comment about Android (ARM) TV boxes that are probably more powerful than our current G5s by now, and the 25 million Raspberry Pi boards sold (for a price of $35 or less), is the part of the argument that I'm trying to stress. I would like to see MorphOS grow into something widely used by hundreds of thousands, if not millions of people (edit: and to do that we need cheap, widely available hardware), but that is probably NOT what the MorphOS Dev. Team members are aiming for, or really want, which is too bad. I think that most team members rather want just a better OS for their personal use, that will have the basics of software written for it by a few friends and maybe a few more interested alternative OS programmers. Without strong growth and a much larger user base, we will never get all the software we need or want for the next generation of MorphOS, unless the team can also invent a translation layer (or something) that will allow the NG MorphOS to also run Linux software (or Windows software for that matter).

I think we all want the next generation of MorphOS to have an Amiga-like structure (just like our current MorphOS) and appearance, but also want a full library of useful software and games, with an up to date web browser. If the NG MorphOS starts out with almost zero software to run natively, and it can only run old Amiga 68k software through emulation, it will have a very difficult time gaining a user base, or programmers willing to create new software for it.

[ Edited by amigadave 03.03.2019 - 09:12 ]

Quote:

I wouldn't bet on it. They were getting faster for a bit there, and seemed to have levelled off. As far as Android TV box goes, if it can throw A/V over to hardware decoding, then they don't need to be fast.

In fact, it's downright a problem if they are, because it needs beefier power supply, more room for cooling, perhaps even a fan, etc - none of which people want for a TV box.

Quote:

And really low power. Launching anything close to a modern browser on one, for example, is a hideous experience regardless of OS. Once you start doing things like trying to turn one into a desktop-like experience, its limitations show very quickly.

If that wasn't enough, many AArch64 processors, like x86-64 processors, lack big endian mode while the cheap RasPi 2 and 3 models retain compatibility to the ARMeb big endian support. Since AROS developer Dr. Michal Schulz has already gotten a Trance-style JIT working on his RasPi, I can't imagine people being willing to register their OS for pay.

Breaking 68k compatibility is the best opportunity to solve most Morphos problems.

Future Morphos should be hardware independent.

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

@ Samurai_Crow
Quote:
So, why port MorphOS to either ARM or x86-64 at all? Based on your logic, since AROS is free and has been running on these for years, there is clearly no reason anybody would ever (pay to) use it, right?


@ amigadave
Quote:
Firstly, as KennyR pointed out, the ARM hardware you can buy for 35 USD is not currently fast enough to properly run MorphOS and a truly modern web browser either.

Secondly, the 35 USD price is a bit of a mirage. If you plan to compare apples to apples, you need to list the price of a complete Rasperry Pi 3 system that includes a power supply, decent-looking case, and Wi-Fi. The prices for complete kits on the Amazon USA website tend to be closer to 80 USD. You can buy x86-64-based mini PCs with 4GB of memory (vs. 1GB offered by the Pi 3) and a fanless quad-core processor for around 100 USD, which clearly provide far superior performance.

Quote:

Fair points, I guess I am looking further forward, rather than what is available right now. The next iteration of the Raspberry Pi is said to include WiFi, as well as being considerably faster. I was not aware that running any web browser was a "painful" experience on the current versions of the "Pi", but seeing how well MorphOS runs on something as weak as the Efika 5200b, if it only had more RAM, I assumed that MorphOS would run fine on the Raspberry Pi 2 and above.

Getting, and keeping up to date, a modern web browser for MorphOS appears to be something that will always be a difficult challenge for a community as small as ours, no matter what hardware we are running on. The main point of my argument for supporting the AArch64, is that it appears that it is growing faster, has cheaper prices at levels of performance that should be more than adequate to run a lightweight OS like MorphOS, and Android, or iOS don't seem to be as dominant as Windows is on x64, so it might be easier to sway new users to MorphOS on ARM devices, than it will be to gain new users on x64.

We desperately need new users and programmers, if MorphOS is going to grow and survive, specially since we will no doubt lose some users who will not follow the MorphOS Dev. Team to the x64 architecture, because they do not want to leave the Amiga compatibility behind and switch to emulation for running their old Amiga 68k software. I think that we have a better chance for getting new users and programmers by supporting AArch64 and the next Raspberry Pi, than we do by supporting only one or two models of an x64 laptop and desktop.

Raspberry Pi owners are more likely to try out MorphOS, than people who will need to buy (or who already own) a specific laptop or desktop computer, which they can more easily run Windows, and all of it available software and games, or Linux, which is free, and will have infinitely more software than MorphOS NG when it is released.

It is a matter of who the Dev. Team targets as potential new users and programmers, or just trying to satisfy themselves, and the few hundred users we currently have.

The only other alternative I can see working, would be IF the Dev. Team can somehow come up with a way to run all existing Linux software on the NG MorphOS natively, or with a very simple recompile, plus provide all the tools Linux programmers currently have to make new software (or some tools which are even better than what Linux has available), so that we can convert thousands of current Linux users and programmers, and convince them it is worth their money to purchase MorphOS NG, and leave Linux behind.

I think the MorphOS Dev. Team are fantastic programmers, but I don't know if the team is up to the task of accomplishing all of the above goals, and convincing the thousands of Linux users and programmers, to become MorphOS NG users and programmers.

They might be able to convert the many dissatisfied AmigaOS4 users and programmers (which is probably only a couple hundred people), and if the team is really lucky, they might convert a few thousand existing and former Amiga 68k users and programmers, and maybe that is the best we can hope for. All I know is that we must find a way to grow our user base, and find new programmers, or all of this is just a waste of time and effort.

[ Edited by amigadave 04.03.2019 - 08:38 ]

Quote:
The people looking for bargain bin hardware are the least likely to spend any money on software, i.e. niche operating systems and applications that are not free of cost. Of course, that is not to say that you should aim for Ferrari-priced hardware because that might ensure every user has proven to have plenty of disposable income.

Nevertheless, targetting the most price-conscious users seems misguided. People who would rather trade 70% of performance to save 20% on cost are the ones who will spam these forums complaining about anybody daring to charge any amount of money for MorphOS. We have seen them in the past and we would see them a lot more in the future.

Quote:
Linux is also available on the Rasperry Pi.

Technically, even Windows is available for it, albeit a cut down version aimed at embedded uses.

Quote:
MorphOS targets people who are dissatisfied with the other available choices regardless of what hardware they currently own.

Quote:

Not if they are dissatisfied with both Linux and the cut down version of Windows on those cheap devices they already own, so I must disagree with you on this point. If the Dev. Team provides a free demo version with the 30 minute time limit, and a low entry price for the NG MorphOS, I believe that many of those Pi owners will be willing to purchase the superior OS. We will never know for sure, unless the Raspberry Pi is supported at some time in the future.

Quote:

Yes, I agree, but if the Dev. Team ports MorphOS to x64 only, and ignores ARM, or AArch64, or more specifically, does not support the next generation of Raspberry Pi's, they will be missing out on hundreds of thousands, if not millions of potential users, who by the very nature of purchasing a Raspberry Pi, have shown their interest in doing things differently, and/or experimenting with alternate solutions. Which is more likely to happen, a future owner of the next generation of the Raspberry Pi will pay "X" sum of money to purchase MorphOS NG after experiencing a demo version of it, or a potential user/programmer, who is looking for alternatives to Windows and Linux, will purchase a specific model of laptop, or desktop computer, just to try a demo version of MorphOS NG? It seems like a simple question to me, with a very simple answer. That in itself does not make the discussion a open and shut case, as there are many other factors, but I think it makes my argument worth looking into.

Edit: Remember, one of, if not the main reason for the relative success of MorphOS 2.0 and above, was the fact that it runs on hardware that can be found widely, and purchased for very little investment. Some users left, once MorphOS became an OS they had to pay for, but many more opted to buy it. In the World of Amiga NG products, MorphOS would have to be considered a success, even if it is not a success anywhere else.

[ Edited by amigadave 04.03.2019 - 13:41 ]

I'm not saying that there aren't very strong reasons for why the Dev. Team chose x64 as their target hardware, I'm just saying that there are some very strong reasons why they should also be supporting ARM, or more specifically, the next version of the Raspberry Pi. Most developers are going to want some compensation for their coding time spent, and the larger the user base, the more potential for coding income on their projects.

The low end hardware provides more users than medium to high end hardware.

[ Edited by amigadave 04.03.2019 - 13:51 ]

> many AArch64 processors [...] lack big endian mode while the cheap RasPi 2
> and 3 models retain compatibility to the ARMeb big endian support.

All AArch64 processors support BE8 mode, which means big-endian data encoding and little-endian instruction encoding, and all AArch64 processors lack BE32 mode, which means they lack big-endian instruction encoding. The BCM2837 with its Cortex-A53 cores is no exception to that. More info and references:

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7675&start=522

Quote:
"We will never know for sure unless..." is not a good reason to do anything but trivial tasks.

Why not port MorphOS to the C64? It could become a huge hit, and nobody can prove otherwise unless it is being done! I am sure you get my point...

Supporting two separate processor architectures is not trivial by any means and brings severe risks of fragmentation, which would hurt the user experience. Porting the OS is the easy part. You are underestimating how much additional work it is you are asking for.

Quote:
Please carefully consider the following: You have no idea what hardware a future x86-64 version of MorphOS would run on but you have already decided that the choices must be less widely available and substantially more expensive than a rumored hypothetical future Rasperry Pi model that has not been announced by anybody.

You are literally asking people to publicly commit to additional work based on your own guesswork, intuition and feelings, not hard facts.

Quote:

Bold words from a man advocating desktop ARM. Fact remains that an operating system with no software is a dead operating system.

I've installed MorphOS on a mac because it's the cheapest-fastest system that can run Amiga software. That was kind of the point of MorphOS in the first place, to create an Amiga upgrade path.

It would only make sense to break Amiga compatibility if A) MorphOS had developed a software library of equivalent or superior utility, and B) that software would not also be incompatible with "New" MorphOS. Neither are true, unless you limit yourself to pottering about web browsing and listening to MP3. I'm not worried because it'll probably take another ten years if ever.

Frankly when you talk about "hundreds of thousands if not millions" of new users, advocate a nonstandard CPU and oddball hardware, predict new OS feature development at a rate that's never happened before, etc etc you sound exactly like an AmigaOne delusional from 2004. Literally the exact same delusions. A healthy person would observe that none of that stuff came true and then study the reasons why.

[ Edited by khorse 05.03.2019 - 15:05 ]

Quote:
There is absolutely no need to suggest someone is mentally unwell because of a difference of opinion.

I can predict what the next RaspberryPi will be like, not because I am psychic, but because of what the RaspberryPi is: a small, low powered developer and hardware hacking board, meant for applications as diverse as an electronic doorbell display to a wireless print server.

Much of that functionality will be lost if it ever becomes much more powerful than it is. It needs to remain USB bus (or even PoE) powered, so it will always be under 5 watts. It will never gain PCI-e slots or the hardware to use them, or support for SATA, or expandable memory, or external GPUs, or any of the other basic things that desktops really should have. And it needs to remain cheap, because nobody wants to build their own microwave display with a 200 euro ARM board.

At best, it will simply get slightly faster, with a slightly more capable GPU (probably focused towards hardware-accelerated 4k video), get more onboard RAM, and have slightly more throughput on its SD card/USB/network mixed interface. It will never be anything more than a novelty (and very slow) desktop if used in that function, because, and I stress, it simply isn't meant to be.

Hmmm for the architecture choice ... i don't kow but what i would do if i was in the dev team is to create a brand new OS , Opensource is not the way because Morphos rights are still to Ralph Schmidt and the devs have a better interest for working on Morphos if this can help them to buy pizzas or food or simply new hardware devices to work on.
If i was a talented developper i will do a modern Os mostly hardware independant to avoid the rewrite of drivers every days.
So i expect to use a Linux Kernel and my Os will use drivers written for linux by a sort of wrapper(a bit more complex), so we don't need to write Nvidia or Ati or Intel drivers but just a sort of wrapper for Vulcan or Opengl.
For using Ppc software we can imagine that it can be encapsuled inside a box (Abox) using the virtualisation facilities.
Virtualisation can be a chance if it's used in a sensible way.
But i'm not developper, but i know that rewritting drivers is a shame and for supporting a great amount of machines using a linux kernel can be a good option (instead of qwark)
I'm not sure that using linux apps like in aros is very intersting , i prefer native stunning apps even if they use opensource parts, if we are on X64 , javascript Jit or Uae Jit and many programs like Mupen64 or DCemu will be easier to port, think of no more endianess issues and a full opengl support...

[ Edited by acepeg 05.03.2019 - 21:45 ]

> Opensource is not the way because Morphos rights are still to Ralph Schmidt

Only for the parts he actually wrote. And I'm sure he can open source them if he wants to.

> we don't need to write Nvidia or Ati or Intel drivers

ATI is so 2000s ;-)

> using linux apps like in aros

Sounds like some misconception of AROS to me.

Just some Talos news, make of it what you will.
(The last paragraphs are notable)

https://www.talospace.com/2019/03/linux-50.html

https://www.talospace.com/2018/08/making-your-talos-ii-into-power-mac_29.html

> https://www.talospace.com/2018/08/making-your-talos-ii-into-power-mac_29.html

You already commented on this article in that Talos thread half a year ago:
https://morph.zone/modules/newbb_plus/viewtopic.php?forum=11&topic_id=11815&start=178 :-)

Quote:

I agree. Though I use it as my public web and mail server.
And it works well for this.

But for desktop use the requirements are much higher, and this device just can't fulfill them. And it wasn't made for that.


Manfred

Quote:

I was not suggesting that as a "reason to do anything", but simply a statement that we won't know who is more correct, unless the Dev. Team decides at some point in time to support ARM, or specifically the next Raspberry Pi. Please don't read more into what I am writing than the simplest interpretation.

Quote:

I got your point perfectly, but it has nothing to do with what I wrote, see my above explanation.

Quote:

I'm not underestimating anything, and I know that the task of porting MorphOS to a different architecture, plus adding all the other features, such as full 64bit support, memory protection, and SMP support, are HUGE tasks. They are essentially creating a new OS, that is in some ways similar to MorphOS as we know it today. As this discussion progresses, and the few members actually add constructive comments, my perspective changes and grows. I am now more hopeful that the Dev. Team can create a new NG MorphOS that is mostly hardware independent, and cross platform capable, so that even if they do not support ARM at first release, they can possibly add ARM support within a year or two afterwards.

Unfortunately, I don't think that using the Linux kernel is viewed as a desirable solution to the problem of writing drivers, to many, or most of the Dev. Team, but perhaps a different kernel that also includes drivers for many hardware choices can be found and used as a starting point for the NG of MorphOS.

Quote:

I'm not asking anyone to commit to additional work for anything! All I am asking for is intelligent discussion regarding the Pros and Cons of different architecture choices. I stated some Pros, as well as a couple of Cons, for choosing the ARM/AArch64 processor, and for the most part have gotten back just garbage and insults in return, so I will not bother to continue with this discussion, and I won't mind if you close this thread right here. No one appears to have any useful comments to make (other than the very few, by yourself and perhaps one or two others).

Well I am enjoying the conversation. At least the part with rational individuals anyway. The handful of yahoos arguing over their own semantics can please go away and let the adults talk.

Anyway, while I see Amigadaves point with the RPI I would much rather have a machine with more power. In a perfect world we could have a modern, inexpensive and powerful POWER processor but that seems to be more and more difficult and costly everyday.

With that in mind then I believe X86 is the way to go, though it pains me to say so.
I really look forward to taking one of my 2 year old PCs and after purchasing a few inexpensive yet SUPPORTED pieces of hardware such as sound cards and NICS and being able to use our wonderful OS.

> In a perfect world we could have a modern, inexpensive and powerful POWER
> processor but that seems to be more and more difficult and costly everyday.

I find 375 USD for a modern and powerful 3.8 GHz 4-core/16-thread POWER9 processor quite inexpensive, and I don't see how this will become any more difficult or costly.

Quote:

Aren't the boards to put it in starting at 800 USD?