Karlos Quote:

I tend to think of them as "enhanced compatibles", but I don't have a specific term. I use PPC/NG because that's the term everyone uses. Yes, I appreciate the irony given the discussion.

Calling "Amiga NG" what it is not, "Amiga Next Generation", seems oddly appropriate in this day and age. At least it didn't end up that way for propaganda purposes or by cancelling/changing another name. Simply putting the AmigaOS on newer more powerful hardware than dead on the desktop 68k was enough to be considered next generation. It was assumed 64 bit, SMP, memory protection/process isolation and improved security would follow. The primary advantages gained were higher performance CPUs and GPUs though, which even emulation can provide now. There isn't even a standard SIMD unit or FPU with the introduction of the A1222+. Amiga NG seems to be falling back to the 68k Amiga which has been brought up to practically on par hardware performance and features using emulation but that hits a wall short of competitiveness as well.

@matthey

The hardware can be seen as "next generation" (relative to 68K) when you have 64-bit multicore PPC on offer. Just not the platform as a whole, once the OS is factored into the equation.

Last edited by Karlos on 06-Feb-2024 at 03:56 PM.

_________________
Doing stupid things for fun...

@jPV

Where Magellan is outdated compared to other desktops? If you combine it with commodities (amigaos) you can add a lot of options, at least on 68k. I did that in my aros 68k distribution extensively. And I used the configuration (f.e. filetypes) extensively too to support much more file formats by default

Last edited by OlafS25 on 06-Feb-2024 at 05:05 PM.

@matthey

Quote:

matthey wrote: OneTimer1 Quote: AOS4 or MOS are 32 Bit, they could be ported back to 68k and with PiStorm or UAE there should be hardware fast enough and with enough RAM to make it usable. This could have be started 20 years ago but no one* really did it and I see no reasons why it should be done now. There were 68k beta versions of AmigaOS 4 modules. My understanding is that there are still 68k modules that were never converted to PPC (ARexx?).

AREXX and IconEdit, AFAIR.
Quote:

How hard can it be to re-enable 68k support in the code? Is it worth it for emulated 68k systems though?

Maybe it's not worth, since most of the code was converted from 68k assembly to C. So, losing performance.
Quote:

It is possible to have 32 bit and 64 bit code running together. ARM CPU cores can do it except where 32 bit support is removed on newer cores. It's not even that special. I believe it is possible to add 64 bit support to the Amiga while retaining compatibility.

Only if the two wolds are isolated and no resources are shared. So, running the 32-bit world on a sandbox inside the 64-bit.

@jPV

Quote:

jPV wrote: Quote: cdimauro wrote: Most of those things were implemented by subsequent (to 3.1) o.s. versions and/or third-party tools/patches. Ambient could also be ported. In general, you can add what's missing to the Amiga o.s.. Of course everything could be done, but wake me up when someone actually does implement them all. I don't want to waste my time on daydreaming.

Me neither. All of them are hobby/toy o.ses, so there's no future for them, whatever enhancement that they could get.
Quote:

You just can't hack & patch OS3 to, let's say, MorphOS level with all the features currently.

Why not? Is there any technical reason for it?
Quote:

And even if you try, the result will be a fragile house of cards, that is no good for anything more serious.

As I've said, all of them are hobby/toy o.ses, so there's no difference.

However I don't see why it would be fragile. As for this, any technical reason?
Quote:

It might look semi-fine if you just use it occasionally, but it's not for productive daily use. Been there, done that.

Same for all parties...
Quote:

And if we'd strictly compare OS advancement, we'd also leave 3rd party hacks out of the comparsion.

Why? Plenty of them were developed after the demise of Commodore and... they worked-out.

Even AmigaOS 3.5 and 3.9 were hacks, because they used the 3.1 Kickstart as the bases, patched a lot of staff on top of that, and provided several enhancements.
Quote:

Quote: It was even possible to completely replace the Workbench with things like Scalos or Directory Opus / Magellan. Yea... I'm a big Magellan 2 fan and I've been using it as a Workbench replacement for 25 years now on all of my HD equipped Amigas. I have it installed currently on five 68k Amigas and on several... how should I call them... "incremental operating systems"?

Enhancements.
Quote:

Unfortunately it is outdated in too many aspects nowadays to be a contender to, for example, Ambient, which is a more modern take on the same approach. And yes, I do know Magellan 2 is open source now and I also donated to get it open sourced, but nobody has really modernized it yet, but it will conquer the world when someone does it! While I wait for that, I'll keep using what's better and available.

That's ok, but... it could be ported, anyway. Exactly like some applications were ported to PowerPC systems, the opposite can be done. There's nothing that prevents it, beside the time of some passionate.
Quote:

Quote: And yes: my statements are perfectly valid, whatever a blind fanatical like you can claim differently, since they perfectly match with the reality. Something that you can't see because you're using your MorphOS-tinted glasses... "ports or reimplementation of the original o.s." gives me the impression that there haven't been anything new in the new operating systems and they'd be simple clones only, which of course isn't the case. I don't think that I need MorphOS-tinted glasses to get that impression...

I never stated that nothing new was added.

All my arguments and writings were/are about the "NG" definition, which I've strictly linked to the o.s. issue which is still carrying after so long.
Quote:

There are also many problems that don't have solutions under OS3 currently, but have been fixed in MorphOS and OS4,

Which ones?
Quote:

so I'm not accepting your "all problems of the original o.s. are still there". Can you elaborate why you claim that nothing has been improved? Or is that "all" limited to certain types of issues?

Karlos and matt already reported the most important ones.
Quote:

jPV wrote: @Karlos Quote: Karlos wrote: @jPV [quote]Of course everything could be done Thank you. This is all I was looking for, either an example of something fundamentally inaccessible to 3.x/68K that warrants the "next gen" definition, or the recognition that it's just a label. Nobody - well not me anyway - is making the claim that MorphOS, OS4, et al are not significant improvements quite the contrary. They are not, however, generational leaps.

Fine by me, I was more upset about belittling comments about improvements in these new operating systems by the other user. The NG term was just a sideline in the discussion for me.[/quote]
If it's fine for you then you're giving me the same credit which you gave to Karlos, since he reported exactly the same point which I shared before.

How could you be upset with me then?

Karlos Quote:

The hardware can be seen as "next generation" (relative to 68K) when you have 64-bit multicore PPC on offer. Just not the platform as a whole, once the OS is factored into the equation.

CPU multiprocessing was not a priority for the AmigaOne. The original 2002 AmigaOne SE 750CXe CPU did not support multiprocessing and most of the AmigaOne systems that followed did not either.

https://en.wikipedia.org/wiki/PowerPC_7xx Quote:

The 7xx family is also widely used in embedded devices like printers, routers, storage devices, spacecraft, and video game consoles. The 7xx family had its shortcomings, namely lack of SMP support and SIMD capabilities and a relatively weak FPU. Motorola's 74xx range of processors picked up where the 7xx left off.

AmigaOne systems were not multiprocessor (sooner) and were only multicore later (2012 X1000 the first?). The decade before multicore arrived to AmigaOne is longer than the original C= 68k Amiga lifespan. The first PPC/Power processor with multicore was the exotic and expensive POWER4 CPU introduced in 2001. The first multicore PPC may have been the 2005 PPC 970MP but I believe Freescale came out with a dual core CPU around the same time. These were expensive workstation class CPUs though. Multiprocessing with multiple CPUs was possible earlier. Apple had the famous dual processor G5 Macs but again expensive and MorphOS still doesn't support multiprocessing. C= experimented with CPU multiprocessing and it was possible because it existed.

A Lock-Free Multiprocessing OS Kernel (uses 68030 CPUs in 1991)
https://cs.uwaterloo.ca/~brecht/servers/readings-new/massalin91lockfree.pdf

Multiprocessing, supported by the 68k, obviously wasn't the motivation to move to PPC and PPC multicore for the desktop didn't exist yet so also was not the motivation. I believe the motivation was higher performance CPUs and GPUs but there is now more performance difference and age between PPC CPUs and modern CPUs than there was between the 68060 and PPC CPUs when the AmigaOne was introduced. Now 68k emulation on modern desktop CPUs is higher performance than PPC CPUs yet expensive PPC Amiga hardware continues to be produced.

cdimauro Quote:

Only if the two wolds are isolated and no resources are shared. So, running the 32-bit world on a sandbox inside the 64-bit.

A sandbox is not necessary for 32 bit code running on a 64 bit CPU. The 32 bit code can live and execute happily in the low 32 bits of 64 bit address space. Resource sharing is tricky but I don't think impossible. It's not possible with some 32 and 64 bit ISA combos.

Last edited by matthey on 07-Feb-2024 at 12:23 AM.
Last edited by matthey on 06-Feb-2024 at 11:23 PM.

Close to common sense, it's useful to have a common knowledge too.

Let's observe how the VA 500 v2+ is being presented:

Quote:

Vampire 500 V2+ is an accelerator board for the Amiga 500. It uses the Apollo core which is a code compatible Motorola M68K processor but is 3 to 4 time faster than the fastest 68060 at time. It also brings Amiga Classic near to Amiga NG by bringing digital video with millions of colours.

Source: here. And I am pretty positive that similar text appeared to some official Vampire/Apollo website in the past and Amiga Store did a mere copy/paste.

In the meantime, it's also useful to judge within historic context: When the NG appeared (as a term) it provided a system (OS) with distinguished a hardware platform that basically offered integrated solution for usb/firewire/ethernet/audio(ahi)/wifi/video(rtg)/ide-sata/plenty of fast ram etc along with latest versions of poseidon/ahi/cgx/mui/file system/new desktop etc. Something unheard of, by that time. While to some extent you could achieve some of the aforementioned features on a classic towerized franken-system (with a G-REX + PPC + several PCI cards combo - btw I am turning the blind eye on the cost factor) it would relied either on old/third-party stacks (drivers) or previous hw chipset versions and ultimately would deliver an inferior experience due to facing several bottlenecks starting from previous CPU generations (no pun intended), crippled bus speeds all over the place and outdated OS 3.x which had great percentage of crashing if you sneezed next to it.

Fast forward 20+ yrs and here we are having ecosystems capable of running the latest WebKit core, doing 4k on PCI-E gfx, able to mix all sort of 68k/wrap/ppc libs/classes/gui, both constantly improving their offerings in terms of hw and available sw but hey, why not wait the 68k scene which in the meantime can't believe a fully functional scummvm port?

Last edited by Cool_amigaN on 07-Feb-2024 at 01:00 PM.

_________________

@Cool_amigaN

Quote:

ward 20+ yrs and here we are having ecosystems capable of running the latest WebKit core, doing 4k on PCI-E gfx, able to mix all sort of 68k/wrap/ppc libs/classes/gui, both constantly improving their offerings in terms of hw and available sw but hey, why not wait the 68k scene which in the meantime can't believe a fully functional scummvm port?

Two points; 1) Only MorphOS/AROS can currently claim functional Web Browsers for modern use. 2) Who cares if you can emulate. OS4 classic or eventually run MorphOS on a x86-64 which you can dual boot into a modern OS?! 3) IS any of this as fun as 68k?

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@Cool_amigaN

...sorry that was three points!

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@Cool_amigaN

Quote:

Fast forward 20+ yrs and here we are having ecosystems capable of running the latest WebKit core, doing 4k on PCI-E gfx, able to mix all sort of 68k/wrap/ppc libs/classes/gui, both constantly improving their offerings in terms of hw and available sw...

No way you wrote all that with a straight face. How many new hardware released in that last 20+ years? How many major OS updates?

_________________
Doing stupid things for fun...

@matthey

Quote:

matthey wrote: cdimauro Quote: Only if the two wolds are isolated and no resources are shared. So, running the 32-bit world on a sandbox inside the 64-bit. A sandbox is not necessary for 32 bit code running on a 64 bit CPU. The 32 bit code can live and execute happily in the low 32 bits of 64 bit address space. Resource sharing is tricky but I don't think impossible. It's not possible with some 32 and 64 bit ISA combos.

The problem is exactly resource sharing: it's practically impossible from the 64 bit world to the 32 bit one.

@Cool_amigaN

you propably refer to MorphOS. But then, why it has not conquered the world if it was that great? Even AmigaOS has more users despite using MorphOS is much cheaper because it uses old used hardware. I say you one secret... amiga users for a long time do not expect to get a modern platform useable in todays terms, it is hobby and fun and not to forget emotion/memories. For many of the current community it got too modern and thus not emotional enough anymore, for people outside it basically miss most what they expect, including the software, drivers, hardware support and so on.

NG 68k hardware today exactly fits in there, it is a kind "what could have been" and memories. So it is not relevant how good MorphOS runs on a used PPC Mac and if it is better than V2/V4 or not (similar to PiStorm)

cdimauro Quote:

The problem is exactly resource sharing: it's practically impossible from the 64 bit world to the 32 bit one.

With 68k address registers, the upper 32 bits of the address are extended. The resulting 32 bit address then exists in the lower 32 bits of the 64 bit address space. A function call to allocate memory could have a MEMF_64b flag that allocates and returns 64 bit address space memory but otherwise returns lower 32 bit memory. Any 32 bit pointers would still work with 32 bit allocated memory on a 64 bit CPU. Most function args would fit in the lower 32 bits of data registers and exceptions could receive 64 bit versions of the functions. For compatibility, it would be good to keep most of the AmigaOS in 31 bit addressable space and this has a performance and memory saving advantage as most OS related pointers in memory could remain 32 bits. With 64 bit capable memory allocation functions and a 64 bit capable scatter loader, hunks that are marked with MEMF_64b could be loaded into 64 bit address space. This is an oversimplification but the changes to the AmigaOS to allow large code and data into 64 bit addressing space for 64 bit aware programs is not too high with the right 32 and 64 bit ISA combo.

MEMF_32b // returned memory can be from upper 2GiB of 32 bit 4GiB address space
MEMF_64b // returned memory can be from anywhere in 64 bit address space

A single mode 32 and 64 bit 68k ISA combo like used for the AC68080 has advantages for enabling 64 bit in a compatible way as mentioned above. The 32 bit code remains compatible and 64 bit instructions can be used as needed. There are several disadvantages to a single mode with most other architectures choosing to create a separate 64 bit mode. Branching from 32 bit to 64 bit code or from 64 bit to 32 bit code requires a mode change and possibly expensive pipeline flush. The CPU has to know where the mode change occurs with several possible indication methods. Thoughts?

OlafS25 Quote:

you probably refer to MorphOS. But then, why it has not conquered the world if it was that great? Even AmigaOS has more users despite using MorphOS is much cheaper because it uses old used hardware. I say you one secret... amiga users for a long time do not expect to get a modern platform useable in todays terms, it is hobby and fun and not to forget emotion/memories. For many of the current community it got too modern and thus not emotional enough anymore, for people outside it basically miss most what they expect, including the software, drivers, hardware support and so on. NG 68k hardware today exactly fits in there, it is a kind "what could have been" and memories. So it is not relevant how good MorphOS runs on a used PPC Mac and if it is better than V2/V4 or not (similar to PiStorm)

Amiga NG 68k has a better chance of catching up with so called Amiga NG PPC than Amiga NG PPC has a chance of catching modern desktop systems. The benefits of Amiga NG 68k are more appealing to Amiga enthusiasts and likely the general population because of increased 68k Amiga compatibility and low cost hardware. Emulation of Amiga 68k hardware isn't good enough though. There are no hardware improvements with the emulation of 68k like 64 bit support, SMP, more modern ISA enhancements, improved code density, chipset enhancements and other improved standards. There is no enhancement standardization and there is no developer support for a virtual 68k Amiga system. Even FPGA 68k Amiga hardware is an uphill battle. Adding enhancements is easier than gaining standards acceptance and developer support. Granted, the AC68080 enhancements have been practically by dictate rather than committee with questionable choices made. Even if FPGA hardware was adequate to gain standards acceptance and major developer support, a high performance CPU in FPGA is not competitive either.

@matthey

Quote:

matthey wrote: @agami There were 2 primary problems that I can see for AmigaOS 4 and MorphOS. 1. The NG OS was standardized before it was adequately improved. Software was created and people don't want to lose that.

Where on other platforms the OS informs apps and moves them forwards, here the apps inform the OS and hold it back.
Only on Amiga NG.

Last edited by agami on 09-Feb-2024 at 03:20 AM.

_________________
All the way, with 68k

@matthey

Quote:

matthey wrote: cdimauro Quote: The problem is exactly resource sharing: it's practically impossible from the 64 bit world to the 32 bit one. With 68k address registers, the upper 32 bits of the address are extended. The resulting 32 bit address then exists in the lower 32 bits of the 64 bit address space. A function call to allocate memory could have a MEMF_64b flag that allocates and returns 64 bit address space memory but otherwise returns lower 32 bit memory. Any 32 bit pointers would still work with 32 bit allocated memory on a 64 bit CPU. Most function args would fit in the lower 32 bits of data registers and exceptions could receive 64 bit versions of the functions. For compatibility, it would be good to keep most of the AmigaOS in 31 bit addressable space and this has a performance and memory saving advantage as most OS related pointers in memory could remain 32 bits. With 64 bit capable memory allocation functions and a 64 bit capable scatter loader, hunks that are marked with MEMF_64b could be loaded into 64 bit address space. This is an oversimplification but the changes to the AmigaOS to allow large code and data into 64 bit addressing space for 64 bit aware programs is not too high with the right 32 and 64 bit ISA combo. MEMF_32b // returned memory can be from upper 2GiB of 32 bit 4GiB address space MEMF_64b // returned memory can be from anywhere in 64 bit address space A single mode 32 and 64 bit 68k ISA combo like used for the AC68080 has advantages for enabling 64 bit in a compatible way as mentioned above. The 32 bit code remains compatible and 64 bit instructions can be used as needed. There are several disadvantages to a single mode with most other architectures choosing to create a separate 64 bit mode. Branching from 32 bit to 64 bit code or from 64 bit to 32 bit code requires a mode change and possibly expensive pipeline flush. The CPU has to know where the mode change occurs with several possible indication methods. Thoughts?

The main problem is not about how you implement the execution of 32 and 64 bit code at the same time, but purely about sharing resources between the two worlds, and specifically about resources allocated by 64 bit applications.

Yes, one solution which I was thing about time ago is about limiting the Amiga o.s. structures to 32 bit (so, not only 31 bit. Because we can also use the upper 2GB of memory for many internal things), but it's not only about them, because we have to think about applications allocating messages and memory, in general, which are shared with other applications.

To solve this issue you have to force such applications as well to continue allocating memory in the 31-bit address space, so making no use of the greater available memory on the new 64-bit system.

It resembles what Apple did with its MacOS Tiger, which was a weird and so less useful solution. In fact, it was completely supplanded by a pure 64-bit solution after that.

So, and again, it's about continuing to cripple the future of the platform. But maybe it's worth a try, since it has no future in the mainstream and it's only a hobby/toy market.

68k is not as fast as ppc.
emulators are emulators and don't count.
68k is nothing special it is cpu like many others.
no reasons to swith to it when you may just use hundred times faster ppc amiga.
my amiga ppc works like my amiga 1200 only 1000 times faster.
I never made graphics in less than 256 colors on amiga and don't need chipset.
in 256 or more colors on amiga from commodore everything is made by cpu.
any real 32 bit big endian cpu is ok.

@matthey


I think that

You talk about stuff that you not own

You talk about stuff that you never used

You talk about stuff where you not understand the details

You talk about stuff where you have no knowledge about.

You talk about stuff where you completely mix up information



What do you think does it help : if you rant about topics where you have no knowledge


Matthey I have offered you per PM to give you information.
To help you to get some knowledge about the topics you speak.

You ignored this offer.
Why do you prefer to not get knowledge to the topics you talk about?

Why do you prefer to talk BS?

@matthey

Quote:

matthey wrote: Amiga NG 68k has a better chance of catching up with so called Amiga NG PPC than Amiga NG PPC has a chance of catching modern desktop systems.

Since noone is actually on neither goals (lots of people talking, not much walking) that isn't even an academic question.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@ppcamiga1

Quote:

ppcamiga1 wrote: PPC is not as available as ARM. PPC is nothing special it is cpu like many others. no reasons to swith to it when you may just use hundred times faster PiStorm powered amiga. my PPC works nothing like my amiga 1200 is 1000 times more expensive and has to emulate anything that ran fine on amiga 1200. I never made anything on amiga so don't need chipset. in 256 or more colors on amiga from commodore everything is possible with HAM.. ARM32 BE is like any real 32 bit big endian cpu, so ok. I am a monkey's bumhole.

There, I fixed that for you.

Last edited by Karlos on 10-Feb-2024 at 11:07 AM.

_________________
Doing stupid things for fun...

@BigD

Quote:

Two points; 1) Only MorphOS/AROS can currently claim functional Web Browsers for modern use

Wrong, Odyssey on AROS is based on the same version present on OS4, which is also superior

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6