As many of you know, I have long supported all flavors of the "Amiga Experience", which I describe as anything Commodore produced and sold, from the very first A1000 running AmigaOS1.x to the last official 68k AmigaOS at version 3.9 (plus any enhancements, such as the official & unofficial Boing Bags), that came later. The Amiga Experience (for me) also includes AROS (though I have very little experience with it, and need to learn more about it), MorphOS (which I have been an active promoter of), and AmigaOS4.x, which I purchased a First Contact AmigaOne X1000 to run it on. I also supported the failed project called Natami, and promoted the semi-finished FPGA Arcade Replay, which will hopefully someday be ready for mass production and sales.

Although I have far less experience than many other users, I have been around quite a while, and had (still have), an interest in all things related to, or inspired by the original Amiga. Lately though, I have been thinking about what the future of the Amiga should be, and wondering if all the attempts at making a NG Amiga have been a huge waste of time and money. If you look at how far we have come, and what any of the NG Amiga inspired systems is capable of, most rational people would agree that my wondering is justified, and I think that 99% of the population would also agree that all NG Amiga systems have been a wasted effort.

This does not mean that I will instantly abandon all of my NG Amiga systems. It does not mean that I won't support future development of AROS, MorphOS, or AmigaOS4. It does mean that my priorities are shifting, and my thinking is that Amiga 68k, or "Classic" Amiga, is what it has always been, and perhaps what it should always remain, instead of any attempt to make it a replacement for Windows, MacOSX, or Linux. That does not mean that adding new features to improve 68k AmigaOS should not be attempted by the remaining Amiga users and developers who still enjoy using and playing with Amiga computers. I strongly support continued work on 68k AmigaOS, by any developer who is interested in doing so. I am very interested to see what the Apollo Team can make possible, if/when they finally release new FPGA based hardware for running 68k AmigaOS, and I would also support the rebirth, or restarting of the Natami project, including the creation of SAGA, or any other improvements to AmigaOS for 68k Motorola CPU's & custom Amiga chipsets, or their equivalent created inside FPGA chips.

The future, of what is now only a hobby, for a few thousand Amiga users, will remain very splintered and varied, but for me, my focus will be aimed more toward enjoying what I already have, and following just how far the original idea of an Amiga can go forward, with projects like the Apollo Team accelerators, and the Natami (if it is ever started up again and brought to a finished state that is available for the public to purchase). What ever the future of Amiga is for you, I wish for it to satisfy all of your hopes and dreams for what it should be (or what you thought it should be). I'll never speak against any of the NG Amiga inspired systems, but I invite you to examine and evaluate how far we have come, where we are headed, and why you are still interested in anything related to the Amiga. Maybe some of you will come to the same conclusion as I have (and many users before me I am sure), that the original Amiga design, even with all of it's faults and problems and short comings, is still the best, for what it was, and continues to be. Maybe it should remain that way, with only incremental improvements as they become possible, not a reinvention, or some difficult to accomplish task of trying to make it more like other mainstream operating systems? Food for thought.

Sorry for such a long post, but I thought sharing these thoughts might be useful to some Amiga users.

Last edited by amigadave on 30-May-2015 at 06:52 AM.

_________________
Amiga! The computer that inspired so many, to accomplish so much, but has ended up in the hands of . . . . . . . . . .

@amigadave

I hope you don't mind, but I think I'd like to go on wasting my time. I require for my use and enjoyment a faster, more capable AmigaOS and computer than Commodore left off with in 1994. Thankfully, I have my X1000 and a developing OS and selection of applications, both new and old, to run on it. Am thankful some people and companies have wasted time and money so I can keep on keeping on with this Amiga thing.

_________________

@mbrantley

As I stated, I am not abandoning my MorphOS machines, or my X1000, and I will continue to run PPC native software on both of those systems into the future. I have no ill will toward any company that continues to promote NG Amiga systems, but MY focus in the future will be on faster and better 68k Amiga systems. The benchmarks released by the members of the Apollo Team are quite impressive, and there is a good chance that we will soon have FPGA systems running 68k AmigaOS at speeds that are close to, or possibly even faster, than similar software running on a SAM440, or SAM460.

My thinking is that none of the NG Amiga systems are ever going to be able to compete with modern Windows, MacOSX, or Linux systems, so I would rather go backward to what the Amiga was originally and enjoy it for what it is and what it can do, and use Windows or MacOSX, or Linux for modern computer needs. That way I am no longer disappointed with what my NG Amiga can't do and my level of satisfaction goes way UP!

Improvements in FPGA performance and the recreation of Amiga chipset and CPU code are now making significant improvements possible, and that is where I plan to expend most of my Amiga efforts in the future. I don't mind, or care that some others continue to work on and hope for success on any of the "NG" Amiga flavors, it just will not be me any longer. I think that the near future will show that most things that PPC Amiga alternatives can do right now, will also be possible on FPGA based Amiga alternatives.

Sorry I won't be promoting NG Amiga systems any longer, but my head and heart just aren't looking at the future the same way any longer. By all means, continue to enjoy your X1000 and AmigaOS4.x. I'll be using my X1000 long into the future as well, but my expectations have gone to ZERO on that system, so I can no longer be disappointed by anything that happens (or does not happen) with AmigaOS4.x, MorphOS, or AROS.

Edit: You mention the speed of an Amiga from 1994, but what about an Amiga with the same or slightly improved custom chipset, that is 10 to 30 times faster than the fastest over-clocked Motorola 68060 Amiga accelerator? That kind of performance opens the door to many new possibilities for 68k AmigaOS, which I think will be interesting to explore. Frankly, even though I think MorphOS is an amazing bit of work, and AmigaOS4.x is getting better every month, they are still both disappointments and so far behind modern computing, I no longer see a need, or desire to push in that direction. AmigaOS on 68k has certain advantages in simplicity and elegance, and as long as I enjoy it for what it is and not expect it to compete with modern OSes or systems, it fills my needs as a hobby OS. The only reason I can see for AmigaOS4.x, MorphOS, or AROS being used, is if some killer application is developed that can only be run on one of those OSes and no where else, and that is not likely to happen. Amiga software runs best on original Amiga hardware, or under emulation, and that emulation can be software only, like WinUAE, or in hardware, like FPGA. All NG Amiga systems have some kind of compromise when running original Amiga software, and the amount of native PPC software is negligible and easily replaced with better alternatives on modern computers.

I don't want this thread to turn into a debate on which choices are best. I just wanted to share my thoughts and reasons for switching focus. I apologize for the "wasted time, effort & money" opinions, as I should have known that they would upset some users. Each user must make their own choices for the future.

Last edited by amigadave on 30-May-2015 at 08:24 AM.
Last edited by amigadave on 30-May-2015 at 08:20 AM.

_________________
Amiga! The computer that inspired so many, to accomplish so much, but has ended up in the hands of . . . . . . . . . .

@amigadave
I am preparing a Attack on Ubuntu with aeros ; )
At least on arm devices

Compare aeros with Ubuntu (from a Linux user point of view)
Aeros/ Ubuntu
- Debian based. /. Debian based
- aros binary execution / not out of the box
- i386 execution faster than qemu / not on Ubuntu out of the box
- wine preconfigured (minus nasty ms libraries)/ not out of the box
- amicloud / Ubuntu stopped the own cloud service
- integrated AppStore / Ubuntu software enter
- new vector based movie coded / only if I allow

In the next months we will add Notification Center, new app launcher, and more.
Than the attack will begin : )

also I changed my mind and aeros will become noobs compatible (a image for raspberry which allows user to chose different systems and installs them automatically).

(RiscOS could be a victim as well ; ))

So from 0.00035% of raspberry pi users we may jump to 1% or even more ; )

There is no waste. Every bit someone adds to any NG brings us a step closer to the Amiga singularity ... A moment where at least one NG system can compete or even surpass at least one Linux distribution : p

Last edited by phoenixkonsole on 30-May-2015 at 08:21 AM.
Last edited by phoenixkonsole on 30-May-2015 at 08:18 AM.

_________________
AROS Broadway - AEROS - Aminux - AmiCloud - indieGO! Appstore - AmiWallet - VAN lossless video codec - AMC Amiga media Center -KrypUnite - LibertyNet - MinX - amigaNX

I never thought I'd see the day I would agree with anything AmigaDave says, but here you go.

I think he's as right as it gets.

Just a note: "68k" systems an order of magnitude faster than a 68060 exist today: UAE boxes. That's where my heart will remain

... until one of the projects AD mentioned really comes to completion.

@amigadave: I report my thoughts as well. First, I define Amiga = ... Amiga! Our beloved 68K machines. Post-Amiga = AROS, MorphOS and AmigaOS4.

Honestly, for Amigas I see (Win)UAE as the only viable option. Today with a PC you can get cheap hardware which makes the Amiga hardware emulation feasible at very good or even impressive speed.
Consider also that currently there's no x64 JIT, which can/will bring a big push on performance when it's done.
However for the future the best solution for running Amiga applications (so, under the o.s.) at top speed is a new multi-core Amiga emulator where:
- one core is used solely for the chipset & RTG emulation;
- one core is devoted to execute only 68K code (JITed);
- optionally one core can be used only for RTG workloads (handling all RTG requests);
- to be seen if the JITer can be run alone on another core.
Other enhanced can be made, for example avoiding emulating a complete RTG card, and directly dispatching RTG operations to the host o.s.: this will increase a lot graphic operations.

Regarding OS4, it sticks to a diying platform. There'll be no new PowerPCs micro-architectures in future, so the existing ones will be recycled. So don't expect big increase of performance here, unless for using more cores; anyway, single-thread performance will be bound to the current micro-architectures.
Hyperion has shown no plan for a better future: they will continue to recycle the patched/enhanced Amiga o.s. 3.1, and will be bound to the PowerPCs.
A possible solution for continuing to survive in this niche is the enhancement of WinUAE, which is now able to emulate PowerPCs at decent speed ("integer" code). Required: some patches to the o.s. to enable more than 128MB of memory used, and to dispatch to the host o.s. the RTG calls (see Amiga/68K above).
Another solution can be to directly go for using QEMU, which eliminates the unneeded old chipset emulation. In the meanwhile, QEMU performance can increase A LOT, especially for FPU (which currently isn't JIT-ed) and Altivec (not supported at all, but it can be, and even JITed).

MorphOS. Well, they already announced the move to x64, so for sure this o.s. will have a future. However the big problem with the post-Amiga o.ses is represented by the lack of modern features (memory protection, resource tracking, SMP). If they will fill the gap, well, I can see a much brighter future for it.

AROS. Similar to MorphOS, but:
- it's already running on x86 and x64;
- is able to address more than 2GB of memory (especially on x64);
- it supports Gallium for graphic cards (no emulation);
- WinUAE 3.0 is being ported to it, so it means that it'll be even able to transparently emulate OS4 on a sandbox ;
- it's open source, so it has one more chance to survive.
However it:
- still misses the 1.0 goals;
- has the same post-Amiga o.ses problems.

In general, I think that for the future a new (Post-)Amiga o.s.-inspired is need. Modern features should be introduced, keeping what is possible from the old Amiga o.s. platform.

The Amiga was a beautiful experience, but we cannot stick anymore with a badly designed o.s.. Mistakes has to be fixed. Wanted features has to be introduced. For a new, modern, solid Amiga experience.

@amigadave
late insight...

@cdimauro

Quote:

one core is used solely for the chipset & RTG emulation;

Why?

Quote:

one core is devoted to execute only 68K code (JITed);

Why?

Quote:

Other enhanced can be made, for example avoiding emulating a complete RTG card, and directly dispatching RTG operations to the host o.s.: this will increase a lot graphic operations.

Did you even try UAEGFX?


Sometimes, I find your answers strange. WinUAE is more than capable of all what you need: fast CPU emulation (if you don´t require MMU), fast RTG GFX, 16 bit sound etc.

Quote:

Another solution can be to directly go for using QEMU, which eliminates the unneeded old chipset emulation

More RAM, USB and faster LAN as a bonus, why not.

Quote:

Well, they already announced the move to x64

Really?


Now my wild dream solution:
Amithlon-like emulation layer for OS4 - CPU emulated by QEMU, basic I/O drivers in host OS (eg. minimal Linux distribution), PCI/PCIe hardware accessed directly by guest OS (mainly GFX card: compositing and 3D).
I bet this would require far lower budget than Cyrus.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: one core is used solely for the chipset & RTG emulation; Why? Quote: one core is devoted to execute only 68K code (JITed); Why?

In both cases: performance. See below.
Quote:

Quote: Other enhanced can be made, for example avoiding emulating a complete RTG card, and directly dispatching RTG operations to the host o.s.: this will increase a lot graphic operations. Did you even try UAEGFX?

From very long time, yes.

I remember that it was software rendered. But I don't know if the implementation changed and now offloads the graphic operations to the host o.s.. If it happens now, well, one point less to be made.
Quote:

Sometimes, I find your answers strange. WinUAE is more than capable of all what you need: fast CPU emulation (if you don´t require MMU),

It's not fast enough, because using one core the CPU has to interleave between chipset AND CPU emulation.

If, instead, you have a single core which (roughly) only executes 68K (translated) code, it'll give a big boost. But it requires a different implementation, so a lot of effort (especially if using some very weird ideas to speed-up the execution )
Quote:

fast RTG GFX, 16 bit sound etc.

See above. I haven't seen the concrete implementation, so I don't know what and how much hardware acceleration, provided by the host o.s., is used.
Quote:

Quote: Well, they already announced the move to x64 Really?

You missed some thread.
Quote:

Now my wild dream solution: Amithlon-like emulation layer for OS4 - CPU emulated by QEMU, basic I/O drivers in host OS (eg. minimal Linux distribution), PCI/PCIe hardware accessed directly by guest OS (mainly GFX card: compositing and 3D). I bet this would require far lower budget than Cyrus.

Absolutely.

BTW, the same approach can be used for the Amiga (68K) platform. If you stick with applications + RTG, you don't need to carry the old chipset infrastructure. You don't even need the (entire) original o.s.: "just" a bootstrap to setup an already configured Amiga o.s. "environment", where old libraries and applications can be transparently executed using a JIT.

For both scenarios I recommend an x64 CPU which supports the MOVBE instructions, in order to:
- use 16 general purpose registers. It increases 68K registers spilling, for much better performance;
- much faster big-to-little endian conversion (and viceversa);
- possibility to use >4GB of memory.

The last is useful for:
- creating large RAM and/or RAD: devices;
- filesystem blocks caching;
- large 68K->x64 translated memory blocks.

@cdimauro

Quote:

It's not fast enough, because using one core the CPU has to interleave between chipset AND CPU emulation.

If you think so...

Quote:

See above. I haven't seen the concrete implementation, so I don't know what and how much hardware acceleration, provided by the host o.s., is used.

Go to EAB and read some WinUAE threads (or ask Toni Willen directly), until then, discussing this topic with you is... waste of time.

Quote:

You missed some thread.

Really?

To be fair, there was big announcement on OSNews about port to new CPU architecture based on posts of one MorphOS developer on Morphzone. However, MorphOS Team didn´t choose new CPU ISA yet (or announced such decision). There is still strong push to other non "x86" CPU architecture like ARM.

Quote:

BTW, the same approach can be used for the Amiga (68K) platform.

Sure, but OS support is rather limited there. Improved AROS 68k could be answer in such case. Then, I wonder, why use AROS 68k emulated on "x86" hardware, when native AROS would offer much better performance...

@amigadave

You are including AROS there but the situation is a little different there. There is also a 68k branch that even runs on average amiga hardware without needing PPC. Wawa and me are right now working on a version for real hardware, there is a option to run it directly from a stick (Phoenixkonsole) and much more. The 68k branch is part of the AROS family and automatically benefitting from progress there. For me personally it is the best of both worlds, on one hand 68k and running most old software and at the same time being in development getting new features and software.

Also interesting is AROS for Raspberry 2 native and of course all related work by Phoneixkonsole (aeros). Raspberry is potentially the perfect platform for AROS and offers the potential to win new users outside existing Community.

Bur in general I share your view, the idea of "NG" to bring users and developers has failed, in my view it already failed before it started because both MorphOS and AmigaOS were too late already. 68k has a magnitudes higher codebase and that has not changed over the years so bringing better hardware to the 68k codebase is more promising than trying to make AmigaOS, MorphOS or AROS to compete with Windows, Linux or MacOS.

@pavlor

perhaps because of software?

I have 10x as much compiler (without counting the former commercial ones)
most applications are for 68k still, there are only a handful really new ones that are "native" PPC
Aros proves that almost snything could be for 68k too, that some are exclusive for MorphOS or AmigaOS 4.X has only to do with programmers not interested in 68k but not real a technical issue

I do not think that 68k can ever be fully replaced (at least for Amigans). It is different for people outside who have no real connections to the old platform so for them it would be just a new alternative platform. They would not miss the old software because they never knew them.

Last edited by OlafS25 on 30-May-2015 at 10:44 AM.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: It's not fast enough, because using one core the CPU has to interleave between chipset AND CPU emulation.

How do you think that the chipset and CPU emulations are realized on an Amiga emulator? By a continuous interleaving of chipset and CPU emulation: the chipset is emulated for some time, then some CPU instructions is are processed, then back to the chipset, and so on.
Quote:

If you think so...

Absolutely. It's the solution normally implemented for an emulator.
Quote:

Quote: See above. I haven't seen the concrete implementation, so I don't know what and how much hardware acceleration, provided by the host o.s., is used. Go to EAB and read some WinUAE threads (or ask Toni Willen directly), until then, discussing this topic with you is... waste of time.

I prefer to take a look at the UAEGFX source, when I've some time, because it's interesting to have an overview of the implementation.
Quote:

Quote: You missed some thread. Really? To be fair, there was big announcement on OSNews about port to new CPU architecture based on posts of one MorphOS developer on Morphzone. However, MorphOS Team didn´t choose new CPU ISA yet (or announced such decision). There is still strong push to other non "x86" CPU architecture like ARM.

Here is the source:

"When MorphOS gets ported to AMD64"

I think that it's clear enough.
Quote:

Quote: BTW, the same approach can be used for the Amiga (68K) platform. Sure, but OS support is rather limited there. Improved AROS 68k could be answer in such case.

Yes, it is.
Quote:

Then, I wonder, why use AROS 68k emulated on "x86" hardware, when native AROS would offer much better performance...

I think that the only reason is to use 68K applications which aren't available otherwise.

@OlafS25

Trance/Petunia-like emulation + RunInUAE would solve any compatibility problems of AROS.

@cdimauro

Quote:

How do you think that the chipset and CPU emulations are realized on an Amiga emulator? By a continuous interleaving of chipset and CPU emulation: the chipset is emulated for some time, then some CPU instructions is are processed, then back to the chipset, and so on.

I don´t dispute obvious advantages of such solution, I simply think performance gain of such big task is not worth the effort.

Quote:

I prefer to take a look at the UAEGFX source, when I've some time, because it's interesting to have an overview of the implementation.

Choice is yours...

Quote:

I think that it's clear enough.

Isn´t that exactly what I write.

Quote:

I think that the only reason is to use 68K applications which aren't available otherwise.

See my above answer to Olaf.

@cdimauro

most of the software base is 68k and will never be ported because of both lack of developers and many sources not available or written in asm. So in my view a 68k OS that is in development and can run both old software and new software is the only realistic option, and there is only Aros 68k.

To win non-amiga related users you need something else like a OS running on Raspberry 2. Again there is only AROS.

AMD64 is of course the best hardware platform for desktops at the moment and I personal would like to do more there in future (time and real life sometimes getting in the way ) but I think any try to make something directly competing with the big platforms has a problem.

Raspberry would be different there, people certainly use it as a kind of fun platform and do not expect to use it as main plaform. When people use 68k (even on a new FPGA) then it is for fun mainly do. If you use something for fun you do not expect it to compete with Windows. In my view too many (perhaps even many of the OS devs) seem to think that they could bring it back to mainstream but that is impossible. You do not need to compete with Windows because that is impossible but you must find a niche. A niche today can still be very big.

@pavlor

Trance/Petunia would be a solution but needs a lot of work. AROS is crossplatform, so you would need versions for X86, ARM and 64bit (AMD64). I found a project where someone tried it but then the project stopped. You can do anything with enough resources but there is one of the problems.

But even if, you use a OS to run software and not just for booting the OS. Why using "NG" at all. You can port almost anything to 68k, it is just lazyness/missing interest why certain programs are not available on 68k.

BTW even MorphOS will only use UAE for 68k integration in case of ISA change

Last edited by OlafS25 on 30-May-2015 at 11:06 AM.

@pavlor

Quote:

pavlor wrote: @OlafS25 Trance/Petunia-like emulation + RunInUAE would solve any compatibility problems of AROS.

It cannot be done on a little-endian or 64-bit architecture. Forget it.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: How do you think that the chipset and CPU emulations are realized on an Amiga emulator? By a continuous interleaving of chipset and CPU emulation: the chipset is emulated for some time, then some CPU instructions is are processed, then back to the chipset, and so on. I don´t dispute obvious advantages of such solution, I simply think performance gain of such big task is not worth the effort.

Unfortunately, chipset emulation takes a big part of the core, and such interleaving process also. I wouldn't have mentioned it otherwise.

That's if we talk about (Win)UAE, of course.
Quote:

Quote: I think that it's clear enough. Isn´t that exactly what I write.

It is, because bigfoot:
- explicitly stated about AMD64 as the ISA for the MorphOS port. No ARM. No discussions about what ISA to chose;
- there is a port to another architecture in the future of MorphOS. The only missing thing is when, because they are currently busy with some other things (X5000 support AFAIK).

I think bigfoot knows what he said, right? Much better than me or you.
Quote:

I think that the only reason is to use 68K applications which aren't available otherwise.

See my above answer to Olaf.[/quote]
See my answer.

@OlafS25

Quote:

OlafS25 wrote: @cdimauro most of the software base is 68k and will never be ported because of both lack of developers and many sources not available or written in asm. So in my view a 68k OS that is in development and can run both old software and new software is the only realistic option, and there is only Aros 68k.

I partially agree, because there's also WinUAE and an Amithlon-like solution, as I already explained.
Quote:

To win non-amiga related users you need something else like a OS running on Raspberry 2. Again there is only AROS.

It's too limited. Doesn't make sense.
Quote:

AMD64 is of course the best hardware platform for desktops at the moment and I personal would like to do more there in future (time and real life sometimes getting in the way ) but I think any try to make something directly competing with the big platforms has a problem.

Absolutely. That's why it makes sense to push the efforts for supporting the x64 port of AROS.
Quote:

Raspberry would be different there, people certainly use it as a kind of fun platform and do not expect to use it as main plaform. When people use 68k (even on a new FPGA) then it is for fun mainly do. If you use something for fun you do not expect it to compete with Windows. In my view too many (perhaps even many of the OS devs) seem to think that they could bring it back to mainstream but that is impossible. You do not need to compete with Windows because that is impossible but you must find a niche. A niche today can still be very big.

It's better to use a cheap and very powerful PC hardware for all this.

@OlafS25

Quote:

But even if, you use a OS to run software and not just for booting the OS. Why using "NG" at all.

Why use 68k? As I wrote, OS4 built-in 68k emulation offers nearly 100 % compatibility and OS (and its supporting enviroment) needs a lot less work than 3.x or AROS 68k.

Quote:

BTW even MorphOS will only use UAE for 68k integration in case of ISA change

I fear that will be not good decision, but they have not much choices (as you wrote, developing new Trance-like emulation layer is not trivial).