@wawa

Quote:

it has features of an ng system

Least features of any NG systems.

Quote:

it has binary compatibility to amiga software library, but not by means of emulation.

OS compliant applications work in Petunia/Trance (or built in interpretive emulation) like any native application (eg. my favourite picture viewer MysticView - 68k application - uses OS4 native support libraries for rendering and OS4 native datatypes for loading of pictures). For games, RunInUAE is fine.

Quote:

aros68k runs without restriction on amiga emulators, allows for almost unlimited ram and rtg options.

Same for my 3.x.

Quote:

aros68k is theoretically able to run on almost any amiga configuration down to 68000 and 2 mb chipram and doesnt need exclusive foreign architecture processor expansions.

OS 3.x is practically able to run on any amiga configuration down to 68000 and 512 kB RAM...

Enough?

@ASiegel

Quote:

You are rewriting history here. Some individuals may believe in this hypothesis but this was certainly not the reason why PowerPC hardware was introduced to the Amiga market way back when. In fact, even Intel were convinced that x86 processors were a deadend and invested billions of dollars in the Itanium CPU line and IA-64 architecture in the mean time. Also, Microsoft had WindowsNT running on Alpha, MIPS as well as PowerPC processors for a time. So, when Amiga Technologies first announced plans to move AmigaOS to the PowerPC architecture, WindowsNT was available for the same CPU architecture.

But Amiga technologies never released a PowerPC port of AOS. It wasn't untill 2000 that Amiga Inc officially announced a PPC AOS port, to be done by Hyperion(which wasnt finished untill 2006-2007, by which time PPC was well and trully dead).

In 2000, the fastest desktop PPC CPU was 500 MHz G4. Intel and AMD were already crossing the 1 GHz barrier. Even then and there PPC was clearly behind(remember, this is exactly why Apple started an x86 Mac OS X project at this time, as Jobs was fed up with Motorola)
By 2000, Alpha, MIPS and PowerPC Windows were fusnotes in history.

_________________

@wawa

Quote:

you put them in quotation, as in being fake, not really amiga users by your definition. its pretty clear.

That is not what I had in mind! Olaf mentioned 68k users as largest part of our community, my point was, most of these are interested only in games not in hardware/OS. I used quotation marks to emphasise "68k users" as much broader cathegory than potential AROS68k users.

@WolfToTheMoon

Quote:

But Amiga technologies never released a PowerPC port of AOS.

AmigaOS 3.9 included WarpOS as contribution (2000).

Quote:

By 2000, Alpha, MIPS and PowerPC Windows were fusnotes in history.

I would put this time few years later - G5 (or any other PowerPC) CPUs weren´t able to compete with Pentium M and later Core/2 CPUs in fastest growing Mac market segment for Apple - portable computers.

@OlafS25

Quote:

OlafS25 wrote: @cdimauro Raspberry is not the typical solution for amigans, for that anything available for 68k is better, either emulation on fast hardware or FPGA. Raspberry is for me typical for reaching non-amiga related users and this group is not as interested in 68k emulation as amigans.

Honestly, Olaf, I don't see how a very fragile o.s. could appetite to non-Amiga related users.
Quote:

In my view that was the basic concept error in all "NG" platforms. They ported/reimplemented AmigaOS with all limitations to new hardware

That's why I don't consider any post-Amiga o.s. as "NG". There is nothing new: they are just enhanced ports to other architecture(s) of a very old and fragile o.s..

@pavlor

Quote:

I would put this time few years later - G5 (or any other PowerPC) CPUs weren´t able to compete with Pentium M and later Core/2 CPUs in fastest growing Mac market segment for Apple - portable computers.

G5 wasn't competitive on desktop either, they even had to overclock it to stay somewhat competitive.
PowerPC was succesful up untill 98-99 timeframe, when Pentium III and Athlon K7 emerged - after that, it was bye-bye.

_________________

@Boot_WB

Yes certainly it could

But if have read it correctly both AmigaOS and MorphOS were financed by others at the start so I assume that the companies not invested money to have some fun but to create profitable platforms.

On Aros it certainly always was different.

@pavlor

Why shouldn´t they be interested in a powerful new 68k platform? Nothing makes more use of resources than games and there is no reason why not new ones could be ported. To run some old games with WHDLoad you do not need that of course.

@pavlor

Os 3.X has MESA/Gallium and is still in development? Really?

@cdimauro

I know that you would prefer a real big break and as I understand it this planned for MorphOS at ISA change but my thesis is that it is 10 years too late for that, at least if you want to do it successfully. A big break at the start would have had more chances with a OS that offers all modern features and has a Amiga inspired API making it easier to adapt the software for it. The problem is that most developers (here I mean 3rd party developers) have long left, many already in the 90s (visible when you look at aminet). For 68k integration you can use UAE but where do you get the native adapted software that justifies to use the OS?

When you make a radical break you have a OS that runs 68k software in emulation but no other software. Hardly attracting new users either.

On Raspberry I see no reason why they would not use a amiga-inspired OS even without MP.

@pavlor

Quote:

pavlor wrote: @cdimauro Quote: Please, show me how to use the same approach for an Amiga/like o.s.. You wrote it is impossible, I proved it is possible - at least in theory.

If I speak of apples and you answer about oranges, yes, everything becomes possible.

And just to recap this part of discussion, I report the sentences, so we can precisely rebuild the context and put a stone to this part.

You: "Trance/Petunia-like emulation + RunInUAE would solve any compatibility problems of AROS."
Me: "It cannot be done on a little-endian or 64-bit architecture. Forget it."
You: Really? (with the link to Rosetta)

Context: AROS means an Amiga-like o.s.. Trance/Petunia is the emulation layer/JITer which let 68K applications transparently run on MorphOS/AmigaOS4. Rosetta compiles PowerPC applications to Intel, yes, but that happens for Mac OS X applications. Mac OS X do NOT publicly exposes its internal structures, so o.s. calls from a PowerPC can be transparently tunneled to the corresponding/native (x86) ones (like what happens with the Win32 APIs called on an x64 version of Windows).

Result: you CANNOT use the same technology (Rosetta) to archive the same on AROS (which PUBLICLY exposes its structures, like any other Amiga/like o.s.; NO tunneling is possible).

So, I report my previous sentence: "Please, show me how to use the same approach for an Amiga/like o.s.."

Danke!
Quote:

Quote: Suppose that no JIT is used (pure emulation) Why not?

Because it was just to give some numbers for two elements which we were talking about : the chipset and the CPU. For the CPU its quite simple to do that (and you can take a look at WinUAE's technical JIT documentation to explain how a non-JIT emulator works; it gives some numbers also), so we eliminate one variable from the system. So we can talk only about chipset variable, which was the desired one (remember?).
Quote:

Quote: So, it means more than 600Mcycles aren't enough for a 1.1Ghz SAM to emulate just the chipset. For SAM, there are other factors: eg. Radeon HD cards don´t support low resolutions (320x256 in this case), their users are forced to use muuuch slower 640x480. Improving UAE to support compositing could be solution there.

I don't think that just a simple screen doubling can be so costly on a SAM with a Radeon HD card. Anyway, maybe a test with a 9200 can be made, but I don't think that it'll change the pictures, since it's the emulation of chipset and CPU the massive tasks of UAE.
Quote:

Quote: the ISA for the next MorphOS port is AMD64. If you thinks so...

Ask bigfoot.

@OlafS25

Quote:

Why shouldn´t they be interested in a powerful new 68k platform?

Quote:

Nothing makes more use of resources than games and there is no reason why not new ones could be ported

What Amiga games need anything better than AGA/68030?

Quote:

To run some old games with WHDLoad you do not need that of course.

Yes, most of your "68k users" do not need that.

Quote:

Os 3.X has MESA/Gallium and is still in development? Really?

OS3.x has at least hardware 3D drivers for Permedia2 and Voodoo3/4/5 cards, unlike other 68k OS.

@ASiegel

Quote:

ASiegel wrote: You are rewriting history here. Some individuals may believe in this hypothesis but this was certainly not the reason why PowerPC hardware was introduced to the Amiga market way back when. In fact, even Intel were convinced that x86 processors were a deadend and invested billions of dollars in the Itanium CPU line and IA-64 architecture in the mean time.

You are rewriting history here, as well.

Itanium was designed for servers and HPC systems. The LONG term plan was to move PC users to Itanium WHEN the such technology was affordable AND 64-bit computing (primarily memory addressing) for "desktop" purposes would have been necessary.

Since there was no 64-bit plan for x86, whereas Itanium was a 64-bit (and new) architecture, that was the natural thing to do.

However AFTER that AMD introduced AMD64/x86-64, and ruined Intel's plan.

Last but not least, at that time (BEFORE Mac OS X introduction to the market) Apple decided to abandon PowerPCs and move to x86. MacOS X was already running on x86, and such architecture would have been used. Then a rampant IBM manager proposed the G5 to Jobs, and Apple decided to continue to PowerPCs, waiting for the new PowerPC processor.

Without such "last minute" change, Apple would have abandoned PowerPCs BEFORE Amiga Inc. started the PowerPC port of the Amiga o.s..

@megol

Quote:

megol wrote: @cdimauro Quote: cdimauro wrote: I cannot give a measure here, just some rough estimates. Emulating a 68000 or 68020 processor doesn't require that much. Suppose that no JIT is used (pure emulation), and that on average 100 cycles are needed to execute one 68K instruction, for a 8Mhz 68000 / 1.4 MIPS it takes 140M cycles, whereas a 14Mhz 68020 / 4.8 MIPS requires 480M cycles. Regarding the chipset, it performs 3.5M transactions/s, where a lot of things can happen for every single transaction because you have a lot different components that works at the same time. Here is very difficult to make an estimate. Using dedicated processors for processor and chipset emulation will make better use of cache resources and allow many optimizations in general. It isn't without problems though - synchronization between the cores will be expensive so in some pathological cases it could perhaps run slower than the standard emulator design or at least much slower than the normal case.

Yes, it can happen. It depends on how much frequently the CPU accesses the chipset registers.

For games I think that the normal approach is recommended. However for an RTG Amiga, the multi-core solution provides much better performance, especially if using some trick.
Quote:

The worst case was (IIRC) code that used the same loop to touch memory and chipset registers as one couldn't simply fast-path the detection but would end up with expensive context switches involving both the CPU core and the Chipset core.

This can be avoided with some other trick.

@pavlor

There are a lot of newer games not ported to 68k because of lack of resources. And perhaps there are new games that might be developed in future when new hardware is available who knows.

No proof and no real evidence against me

We will see what happens when there is really new and better hardware for 68k.

@cdimauro

Quote:

If I speak of apples and you answer about oranges, yes, everything becomes possible.

Quote:

And just to recap this part of discussion

You: It cannot be done on a little-endian or 64-bit architecture. Forget it.
Me: Rosetta?

Done on little-endian or 64-bit architecture.

Quote:

I don't think that just a simple screen doubling can be so costly on a SAM with a Radeon HD card.

I see your experience with emulators is rather limited. That explains your "answers". In UAE, you need far more CPU resources for hires/double lines configuration (in comparison to lowres/single lines). As even you wrote, chipset emulation is CPU intensive, for highres/double lines mode, you need 2 times more CPU power in both x and y directions. That may be not problem on faster G4 CPUs, but is too much for 440/460. If you have access to slower hardware (eg. faster Pentium II or slower Pentium III class), you may try it yourself.

Quote:

Ask bigfoot.

I will rather wait for official announcement.

@OlafS25

Quote:

There are a lot of newer games not ported to 68k because of lack of resources. And perhaps there are new games that might be developed in future when new hardware is available who knows.

WHO will port them?

@pavlor

Quote:

Quote: it has features of an ng system Least features of any NG systems.

what features does aros lack in comparison to os4 as example? gallium? multicore support? cheap and widely available hardware? legacy hardware support? working web browser? low memory footprint?
im pretty sure aros in uae is more usable than os4 and has more features.

Quote:

Quote: it has binary compatibility to amiga software library, but not by means of emulation. OS compliant applications work in Petunia/Trance (or built in interpretive emulation) like any native application (eg. my favourite picture viewer MysticView - 68k application - uses OS4 native support libraries for rendering and OS4 native datatypes for loading of pictures). For games, RunInUAE is fine.

petunia/trance is cpu emulation. you cant run os4 application on amiga without ppc accelerator no matter what.

Quote:

Quote: aros68k runs without restriction on amiga emulators, allows for almost unlimited ram and rtg options. Same for my 3.x.

yes, also in this respect 3.x is superior to os4. i like 3.x and it has still some advantages also in comparison to aros. one of the most obvious is speed and still better hardware support. but i want to have an option for improvement. if this option has not been blocked on os3 by certain entities our situation would be definitely better. aros is an answer to this problem.

Quote:

Quote: aros68k is theoretically able to run on almost any amiga configuration down to 68000 and 2 mb chipram and doesnt need exclusive foreign architecture processor expansions. OS 3.x is practically able to run on any amiga configuration down to 68000 and 512 kB RAM...

same as above. besides the very low end systems may not need aros anyway. so its fine like it is.

@pavlor

Quote:

Quote: There are a lot of newer games not ported to 68k because of lack of resources. And perhaps there are new games that might be developed in future when new hardware is available who knows. WHO will port them?

mazze is updating and including software in aros ports section. same source for all architectures. a lot of that compiles for 68k. the rest eventually will, with much less hassle as doing the whole job on ones own, even using the unholy ixemul library.

@pavlor

In ideal case they just need to be recompiled. In case of applications it already works.