@Karlos

Emu68K It’s playing a low quality 720p video, at low FPS, not that impressed, it need a lot of work, you need YUV420 support, overlay or composite I don’t care, you to expose the native GPU.

You are talking about things I’m not that interested in, so I might be wrong, but I expect you need a complete new gfx driver, to enable 3d and video hardware decoding, I expect you need funding, something you don’t have. This is why you are lobbying for it.

Now Emu68K feels like copy of old Mac_68K_emulator in MacOS7.6.x, it essential enabled PowerPC to emulate 680x0’s to use as replacement for 68K CPU’s

https://en.wikipedia.org/wiki/Mac_68k_emulator

it gave PowerPC bad reputation for being slow due to Emulation.
Then we had MacOS9 that was PPC native finely proved PPC G4 can be fast, at least at the time competed well, of cause did not take long before IBM and Apple screwed it up.

MacOS7/8/9 was held back by legacy, the end result was everything had to be replaced, when they wanted SMP and proper memory protection. Even MacOSX lagged behind.

I’m with Hans on this, back porting to 68K is lot of work, and only degrade what was improved. EmuPPC be a better idea, but only a stop gap solution, sure it provide low cost solution, that is missing, but I expect high-end PPC has place for sometime to come.

with chip shortage, it can be hard to find stores that have raspberry in stock.

Last edited by NutsAboutAmiga on 03-Mar-2023 at 07:31 PM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@ppcamiga1

Quote:

ppcamiga1 rolled is face on the keyboard to emit: pistorm is wortless crap that made amiga nothing more thna docking station for rpi.

No, PiStorm makes the Amiga the thing you keep on saying it always should've been in the first place: A fast, compatible big endian 32-bit processor with a side dish chunky pixels, while remaining backwards compatible in the most immediate and direct way possible.

The PPC is completely worthless today. New hardware costs a fortune and you can only use a fraction of the processing power of it thanks to the inherent limitations of the OS you are putting on it.

The RPi has some of the same problem: AmigaOS 68K is a single core 32 bit system. However that doesn't mean Emu68K itself has to be limited that way. And should it be the case that there's no way to use more than one core for the 68K JIT, it may still be possible to use the other cores for some other native tasks. And if that's not even possible, well it only cost you a fraction of that PPC hardware so no great hardship

_________________
Doing stupid things for fun...

@NutsAboutAmiga

You missed the point by a mile. Emu68K wasn't playing the video, Riva was. A 68K AmigaOS application, completely software decode path, playing 720p at a perfectly watchable rate. As for the quality and framerate you have no way to judge objectively because you can't tell if it's the capture or upload.

None of which matters that's how fast 68K is right now, and it's only going to get faster. The performance gap between NG and classic just experienced a seismic closure. It's only going to close further. There's far more interest in 68K so it's inevitable. There's no putting the genie back in the bottle.

Exposing the video and 3D capabilities of the Pi will surely come just as exposing the framebuffer for RTG and other resources did. Sure the embedded GPU probably won't be as fast as a higher end Radeon but we don't have much software where the GPU is the bottleneck anyway.

Quote:

with chip shortage, it can be hard to find stores that have raspberry in stock

It's the Pi4b and CM4 that are hard to get. You aren't forced to use these if you don't want the fastest possible experience. Many people have them sat round doing next to nothing as it is from before the chip shortage.

And at the end of it all, at least they are still in production.

I may pay over the odds if I want a Pi4 right now but how much over the odds do I pay for a PPC system?

Last edited by Karlos on 03-Mar-2023 at 07:49 PM.
Last edited by Karlos on 03-Mar-2023 at 07:43 PM.

_________________
Doing stupid things for fun...

@NutsAboutAmiga

Quote:

I’m with Hans on this, back porting to 68K is lot of work, and only degrade what was improved. EmuPPC be a better idea, but only a stop gap solution, sure it provide low cost solution, that is missing, but I expect high-end PPC has place for sometime to come.

Show me where Hans says porting OS4 to 68K would "only degrade what was improved." . You can't, because he didn't say it. He pointed out some engineering challenges. I don't deny for as second there won't be any. They surely are not insurmountable. As for emulating PPC, go right ahead. QEMU is there. You can experience it for yourself.

Perhaps you forget, but I was involved with OS4 development in the 4.0/4.1 days. I've seen and worked on the codebase. Unless someone has replaced all the carefully rewritten C code with PPC assembler, there's nothing to "degrade" because no part of it (beyond the lowest level glue and some optimised versions of things like memory copy) is PPC dependent. OS4 is as portable as any "designed for 32-bit big endian" software can be. ppcamiga can actually feel validated by this. If we were talking about porting to a little endian and/or 64 bit platform, there would be real and immediate issues to solve. We are not talking about that. We are talking about recompiling for more widely appreciated and accessible 32-bit big endian platform in 2023: 68K

Even without the PiStorm, I am certain UAE has more users than any PPC Amiga, and I include Macs running MorphOS in that.

Why release 3.2 at all if there's no appetite for 68K?

_________________
Doing stupid things for fun...

@ppcamiga1

Quote:

ppcamiga1 wrote: pistorm is wortless crap that made amiga nothing more thna docking station for rpi. which is stupid because amiga keyboard and mouse where copied from pc 40 years ago.

I mean... Whiskey Tango F^@&ing Foxtrot?

_________________
All the way, with 68k

@AmigaOS4

A port is coming. The people at the helm accept this as a reality. They're only waiting for this blasted A1222 to be completed, and then there's no more PPC hardware for AmigaOS.

I agree with @Karlos, given the recent developments in high performance 68k options, and native environment for legacy software, it would make most sense to port it to 68k. Plus it would kind of make history.

Alas, that gives our industrialist class very little to control, so likely they will see this Emu68 as the basis for the next "Petunia" and port AmigaOS 4 to ARM, with Emu68 providing a compatibility layer.
They can't exploit Amiga users through R Pi sales. But working with Varisys or Acube, they can get users to fork out $500+ for something that is equal in spec to an R Pi 4 board, but with a PCIe slot. That way users can use a nickel's worth of performance from their $200+ GPU.

Last edited by agami on 04-Mar-2023 at 10:39 AM.

_________________
All the way, with 68k

@Karlos

Quote:

@NutsAboutAmiga Quote: I’m with Hans on this, back porting to 68K is lot of work, and only degrade what was improved. EmuPPC be a better idea, but only a stop gap solution, sure it provide low cost solution, that is missing, but I expect high-end PPC has place for sometime to come. Show me where Hans says porting OS4 to 68K would "only degrade what was improved." . You can't, because he didn't say it. He pointed out some engineering challenges. I don't deny for as second there won't be any. They surely are not insurmountable. As for emulating PPC, go right ahead. QEMU is there. You can experience it for yourself.

To be clear, my reason for suggesting adding PPC emulation is because I think it'll be less effort, less messy, and allow existing AmigaOS 4 users to take their current software with them. Plus, it wouldn't require permission from Hyperion. An interesting challenge would be to get OS4 to use Emu68 instead of Petunia for 68K code...

I don't think qemu is designed for top performance, and it might actually be better to take Emu68 as the starting point for EmuPPC.


@agami
Quote:

Alas, that gives our industrialist class very little to control, so likely they will see this Emu68k as the basis for the next "Petunia" and port AmigaOS 4 to ARM, with Emu68k providing a compatibility layer. They can't exploit Amiga users through R Pi sales. But working with Varisys or Acube, they can get users to fork out $500+ for something that is equal in spec to an R Pi 4 board, but with a PCIe slot. That way users can use a nickel's worth of performance from their $200+ GPU.

So cynical... No industrialist would bother building custom ARM boards for a tiny market.

If AmigaOS 4 is ported to ARM (which I hope will happen), then it would make sense to use the built-in GPU. They're more than good enough, unlike the pathetic GPUs/framebuffers built-in to PowerPC chips (like the SM502).

Hans

Last edited by Hans on 04-Mar-2023 at 06:06 AM.
Last edited by Hans on 04-Mar-2023 at 06:04 AM.
Last edited by Hans on 04-Mar-2023 at 06:04 AM.

_________________
http://hdrlab.org.nz/ - Amiga OS 4 projects, programming articles and more. Home of the RadeonHD driver for Amiga OS 4.x project.
https://keasigmadelta.com/ - More of my work.

@Hans

I'm unconvinced about the EmuPPC concept. What software for PPC/OS4 exists that's not in a recompilable state? There can't be that much. It's a lot of work to implement a high performance JIT for the PPC.

Moving directly to ARM overcomes the current hardware deficit and having a high performance ARM replacement for petunia helps the case. However, moving directly to another CPU architecture also perpetuates the need to have multiple end user versions of the same software.

The case for moving to 68K isn't just that suddenly 68K is fast thanks to emulation. 68K running on UAE on modern x64 has already been faster than most PPC native systems for some time. However I do understand that people want to use physical machines for all kinds of reasons.

68K is the Lingua Franca of the Amiga. The only reason we moved to PPC at all was due to the fact the 68K hardware line was effectively killed off and the PPC was the "accepted" solution, largely in part because apple had already done it and we had, for a time, PPC in classic machines teasing how much brighter the PPC future could be.

For a time it was good, but it soon devolved into a case of increasingly niche and expensive and buggy hardware that existed only to run OS4. MorphOS did the right thing there by moving to Mac, for which a sizeable second hand market had appeared. In doing so it neatly sidestepped a lot of the issues that plagued the custom route: Affordable, well engineered and battle tested. Well defined hardware specification to go on.

OS4 had a working Mac port, Moana, but it didn't come to fruition. I don't care what the reasons were any more it was just one error in a long line.

We arrived at the point where apple exited PPC. There's plenty of hardware left for MorphOS, though the pond is drying up there. They have an x86 (x64?) version under wraps but I expect the issue there is one of backwards compatibility. OS4 continued with custom PPC designed that contine to be more niche and expensive.

We are decades into this trajectory now. We could say PPC set an expected level of machine performance for OS4 users and that's a reason not to "go back" to 68K. Except that's a lie. It's a lie because 68K JIT has been performant for a long time now.

We could say there's no hardware 68K solution that's performant enough to "go back". That's also a lie. Amithlon was the exemplar there, albeit unmaintained now. However, thanks to PiStorm there is a fast enough hardware solution. One that is the most compelling of all to an actual Amiga hardware enthusiast: something physical that plugs in and gives their machine performance enough to do what any existing PPC NG systems can do and it's only just getting started. For those that don't want to use a classic machine standalone is an option.

There are multiple ways to run a high performance 68K port of OS4. More ways means more users. More users means more sales. More sales means maybe the deadlock on development can be lifted.

PPC set some unrealistic expectations. 64 bit, multicore SMP. All of which are difficult to do without becoming incompatible. The reason people demand them is in part because they think it's necessary but also because they've had to shell out a fortune on hardware to run an OS that is still 32-bit, single processor centric and leaves their hardware largely unused.

Personally, I'd rather see AmigaOS in its current form become the best, single user 32-bit uniprocessor OS in existence than a half-baked partial jack of some trades, master of none.

SMP is still a tangible goal to chase and perhaps Emu68K can deliver by allowing more than one JIT concurrently by running on more than one core. Even if that's impossible, there are potentially other things that could make use of the additional cores.

However you look at it, 68K is the Amiga's home. It's time to come back.

_________________
Doing stupid things for fun...

Apologies for the OP for the total hijack.

_________________
Doing stupid things for fun...

@Karlos

Yep. 68K is the common base for any Amiga related project: although native code would be the best solution, having a fast 68K virtual machine running legacy code at blazing speeds is still an acceptable solution, moreover if you turn on a machine and you are brought to the 68K environment without noticing the external environment allowing that.

Just look at RGL's The C64 (maxi): you can even turn it on straight to C64 mode and almost not noticing it's a emulated kernal.

This direction has already been written and set, so sometimes it will simply happen. The choice for OS4 developers is simply between being there, or letting AROS 68K taking its place, like it's already happening on Vampire.

@paolone

The dilemma for OS4 is that it ceases to move without sales. Without affordable new hardware there's no sales. Most existing users must have OS4.1 final edition by now.

If you do get a new system, in order properly use that system you also have to buy the enhancer software separately. It's an increasingly poor experience to attract any new users. And where are you going to attract them from? There must be a diminishingly small number of OS4 users that had never used an Amiga previously.

If OS4 is to survive I think it needs to get away from being a PPC exclusive. Moving to ARM native would be a good step but I feel it still creates some of the same problems that moving to PPC caused in the first place. The 68K market is just much larger (for Amiga specifically). Factor in all the emulators out there as well as existing expanded classic machines.

_________________
Doing stupid things for fun...

@Karlos

Quote:

I'm unconvinced about the EmuPPC concept. What software for PPC/OS4 exists that's not in a recompilable state? There can't be that much. It's a lot of work to implement a high performance JIT for the PPC.

To me, whether the software is in a "recompilable state" or not is irrelevant. Herding everyone back to 68K is also a lot of work, with its own set of challenges.

From my perspective, if I'm running the OS on a PiStorm, then sooner or later I'd want to be able to run native code. In that case, porting back to 68K would be an intermediate step to get to native ARM.

Quote:

If you do get a new system, in order properly use that system you also have to buy the enhancer software separately. It's an increasingly poor experience to attract any new users. And where are you going to attract them from? There must be a diminishingly small number of OS4 users that had never used an Amiga previously.

Nitpicking here, but I see no reason for a new AmigaOne machine to be shipped without both the OS and the enhancer (unless specifically requested).

Hans

Last edited by Hans on 04-Mar-2023 at 10:31 AM.

_________________
http://hdrlab.org.nz/ - Amiga OS 4 projects, programming articles and more. Home of the RadeonHD driver for Amiga OS 4.x project.
https://keasigmadelta.com/ - More of my work.

@Hans

Quote:

No industrialist would bother building custom ARM boards for a tiny market.

In early 2009 I could have dismissively said “No industrialist would bother building custom PPC boards for a tiny market”, yet here we are.

Last edited by agami on 04-Mar-2023 at 10:55 AM.

_________________
All the way, with 68k

@Karlos

Quote:

Karlos wrote: @paolone The dilemma for OS4 is that it ceases to move without sales. Without affordable new hardware there's no sales. Most existing users must have OS4.1 final edition by now. The 68K market is just much larger (for Amiga specifically). Factor in all the emulators out there as well as existing expanded classic machines.

The issue is refusing to understand a simple truth: keeping on considering the Amiga fanbase as a 'market' is a self-delusional and self-destructive behaviour. There is NO market anymore and NO external interested people to sell the product to, for the basic simple reason there is NO need for a 30-years old operating system that's just "simple and fast", but lacks of critic base-features like network security, real memory protection, multisuer-capability, SMP and virtualization. AmigaOS, like MorphOS and AROS, are just hobby operating systems serving a progressively vanishing community with almost no way to attract 'foreign' people.

The only chance to survive, and for a limited time before becoming a historical curiosity like dozens more famous ones, is opening everything, let the community bring on the project and accept that the platform is commercially dead. And sell a 68K version in the meanwhile, if still legally and technically possible.

"it's dead, Jim".

@agami

Quote:

agami wrote: @Hans Quote: No industrialist would bother building custom ARM boards for a tiny market. In early 2009 I could have dismissively said “No industrialist would bother building custom PPC boards for a tiny market”, yet here we are.

...and here *you* are with costy, underpowered, customized industrial PPC design adapted to run a OS which neither exploits all the features of the hardware, self convincing that this is the prosecution of the once-glorious Amiga platform by Commodore, just because there is an OS the 'Amiga' name inside and a progressively enhanced recompile/replacement of the old 68K stuff.

PPC has been a mistake. Commercially, technically. It's finally time to accept this.

@all

And let me be clear about this: if an Amiga OS would like to survive, the best hardware it should target is RGL's The A500 Mini.

Just make a deal with RGL and create a proper port, either running in the 68K environment (so it can be run also by classic Amigas and UAE) or straight to the ARM level, with a WHDLOAD frontend to replace the carousel.

@Hans

As I said, moving to ARM wholesale solves the immediate hardware deficit problem but now you are left with the additional problem of providing both high performance 68K and PPC backwards compatibility. That seems like a failure to learn the lessons of moving to PPC.One such lesson being, "native is the path to dissatisfaction by underutilisation".

AmigaOS itself would need to scale up to 64-bit SMP at the very least to fully utilise what Pi can offer. A laudable goal, but if moving to 64-bit SMP were a straightforward task, we'd have pursued that path since the X1000 landed.

Assuming this represents a formidable engineering challenge/dilemma for backwards compatibility then the native performance of the hardware is better put to task in bringing the best possible 32-bit single threaded performance for Emu68K (or EmuPPC). I think SMP in some for is still on the table perhaps and it doesn't feel like a huge leap to imagine a bare metal emulator per core for example.

Even if that's not possible, the available native CPU power can be put to use in other ways, by providing virtual devices for other uses.

I see the "OS should be native" argument as a bit of a moot point. Why? Because most of the time at a computer, especially a uniprocessor one, is spent either doing absolutely nothing, or running application code. What does the OS do that is CPU performance critical rather than say IO latency critical?

Probably the most important performance critical thing any OS actually does is to move data. You need it for actual IO, task switching, interrupt handling and servicing copy requests on behalf of applications. The things that are actually latency critical. Data movement itself is largely bound by memory performance. Do we really think that the emulated code is going to be that much slower at this task than native?

I am of the opinion that a native code passthrough for Emu68K would be beneficial for certain compute heavy workloads that would benefit from native vectorisation (as an example), but I know michalsc disagrees. And his argument is basically the same as my argument as to why we should migrate/return to 68K in the first place so I can't really disagree. There are other ways to skin that cat, one of which I've already made some experimental progress with.

"Herding everyone back", you say. Well, I hate to say it, but how many is "everyone"? There are many more potential 68K customers going untapped then there are wild PPC users to be rounded up and taken against their will. My suspicion is that most wouldn't care to be honest. If Emu68K existed as a bare metal emulation on PPC that could in turn run a pure 68K OS4 experience on their machines with no real perceived loss in performance, I can't see much complaint. No doubt some PPC users would be severely butthurt over "going backwards" to 68K, but I'd wager most of them have no idea what that actually means from an end user perspective. In any case as I've said before I don't mind OS4 being developed for PPC alongside the proposed 68K path, it just shouldn't be PPC exclusive. At least if it wants to survive.

The nitpick is fine but it's still a cost to be included on top of what is already an expensive proposition to begin with.

Hilariously the point was made by someone else about the unavailability of the RPi 4 earlier (desire the fact you can use RPi 3). Someone please remind me how long we've been waiting for Tabor?

Last edited by Karlos on 04-Mar-2023 at 12:06 PM.
Last edited by Karlos on 04-Mar-2023 at 11:48 AM.

_________________
Doing stupid things for fun...

amiga either have to be work or hobby.

If amiga is to be work it compete with win/lnx/osx
and have to be no more than 10 years behind win/lnx/osx.
Which means amiga gui and graphics on top of unix.

If amiga is to be hobby it should be something different than commodity hardware.
Which means no x86, arm or risc-v.

68k emulator on commodity hardware like pistorm is worth nothing shit.
it combine worst flaws from work and hobby.
It is not interesting because it is commodity hardware.
And it is not fast and modern as win/lnx/osx.
(in arm case not fast and modern as lnx/ios).

@ppcamiga1

Shssh. Grown ups are talking.

Amiga is not going to compete with any modern mainstream OS. That doesn't mean you can't use it as your daily system and never has meant that. You aren't going to be able to use it for work, unless your work needs are simple.

So it's a hobbyist platform. That doesn't mean it needs to cost you an absolute arm (no pun intended) and a leg to buy into.

You are obviously fixated on PPC. Well, here's a dose of reality for you. I know loads of people today who remember the Amiga. I know barely any that have heard of the AmigaOne or whole PPC era. PPC has no value today. It's expensive, underutilised and nobody remembers it even existed. Hell, Apple users are all forgetting x64 even existed as we speak.

I know people that bought RPi, sometimes multiple, and half of them are doing nothing. Why? Because they were cheap. Half of the devices could run OS3.x on Emu68K standalone very respectably.

I know the penny never falls for you, but thankfully you are the edge case nobody has to worry about.

_________________
Doing stupid things for fun...

@ppcamiga1

Quote:

68k emulator on commodity hardware like pistorm is worth nothing shit.

Then don't buy 68k/Arm hardware, go back to your PPC shrine and your greased up photo of Ben Hermans but also accept that the majority of people DON'T agree with you and will invest in Arm based hardware WHATEVER you say!

Maybe you'll be silenced when the last PPC embedded controller falls off the production line. It is likely we'll see that in our lifetime, though POWER based chips will likely carry on.

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