@agami

Agree 100% that the PPC hardware is reaching limits.

To me that is a small part of the story. All the interest in new hardware and emulation is great and all... But, without good software and OS features, it almost is just as crippled as it would be on a 100GHz system. Running bad software at light speed! :)

I greatly prefer the one camp focused on the effectiveness of the software so you actually can do something productive with it instead of hacking the heck out of it to maybe get email.

So if MorphOS, for example, stayed with PPC for another 3 years and continued to develop PolyOrga, Iris, Wayfarer, Pagestream, and other software including the OS. I find much more value with this approach. I use it for my work 5 days a week and the 2.5GHz PCIe PowerMac works more than fast for almost everything I do with it. The software has evolved immensely in the last 5 years and are very good.

Probably in the minority here, but I look forward to software improvements much more than hardware ones.

@Matt3k

Software and hardware improvements are far from mutually exclusive. It's great getting software updates as long as you have functioning hardware to run it on. At some point the supply of compatible old PPC hardware is going to run dry.

There's still QEMU, I suppose.

_________________
Doing stupid things for fun...

@Kronos

Quote:

SMP and 64Bit are both nogos inside the current API/ABI and while half a#### attempts may kinda work they would still be a bad decision.

So you're not wrong. The limitations are real enough and as per the general thrust above, if these were simple, compatible upgrades, I'm sure we'd be using them.

What does intrigue me are the "half arsed attempts that kinda work" that become possible to a purely virtual environment and an operating system that is able to leverage the underlying "substrate".

Private memory is one example. If you share it, your application is broken. But in theory, you can have a wedge of allocated memory only for your process that can never be shared but exists at some overlapping logical 32-bit address range with someone else's. Because they are private and must not be shared, there's no reason these can't be mapped to different physical regions that are in completely different address spaces. The net result is not 64-bit, but it does allow applications aware of private memory to use more than is imposed by the logical limit.

SMP via a virtual "hardware intended to simplify the problem" rather than "done properly" might also more practical than current ideas.

Neither of these necessarily break compatibility with real 68K silicon. You just can't use more memory than exists and you just can't run more than one process truly concurrently.

_________________
Doing stupid things for fun...

@Karlos

Quote:

Karlos wrote: Private memory is one example. If you share it, your application is broken. But in theory, you can have a wedge of allocated memory only for your process that can never be shared but exists at some overlapping logical 32-bit address range with someone else's.

Sure, should be simple enough (tm) but what would be the real benefits?

On a system with (near) 2GB available from exec you can have 1 app using almost all of that as long as you don't run any other "heavy" apps.

With such a system (and more RAM) you could do it for more than 1 apps, but each of the apps and the OS would need to keep track of what is what.
Maybe do it in a way where that virtual RAM is always mirrored into the upper 2GB of the 32Bit address space.

All you'd be doing is kicking the can down the road for a few more years till 2+2GB ain't enough anymore at which point you would need a proper solution that also is backwards compatible with the above halt measure...

Since you would need to do some major rewrite of any app utilizing that 2+2 approach, why not go all the way have spawn a proper 64Bit/SMP thread on a minimal kernel (or the one already running the 68k EMU) that you would need anyways to keep track of the 2GB upper memory blocks.

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

@Karlos

Sure, very valid point. Agreed. But at this point of history, software is much more needed than hardward. You can't do anything productive with it. About everything is way to limiting or primitive.

I remember back in the day when dos pc's were the rage. No one cared about the hardware, they bought the things because it did work for them.

So I still stand behind software being more important than hardware overall. I enjoyed play ing with my 3000's hardware recently for one example. Then I tried getting email to work and it was a total joke. Iris comes with a base install of the OS, updates right from the program, and deploys in seconds and works perfectly well for any scenario I throw at it (calendar invites, cancel meetings, multiple email servers, speed and performance and zero drama.).

To achieve this with Iris it took 5 solid years of development and I can use is just as functionally as any other email client on any platform. Creating professional software you can daily drive is no small feat.

QEMU does look interesting btw...

Last edited by Matt3k on 22-Dec-2024 at 06:56 PM.

@Kronos

If you want proper 64 bit and SMP, there's AROS and AxRT. That wins the "NG" promise hands down. Why aren't we all using it all the time? One price you pay is the disconnect and lack of backwards compatibility.

I've come to regard PPC/NG as a folly. It's alone in this weird isolated place. Not quite retro, not current, just extant. Barely.

I don't have any illusions that AmigaOS will be a viable alternative to modern mainstream OS for most people. It appeals to those that like esoteric things or were fond of the original. That's us, old farts knocking on death's door as it is. The best we can possibly hope for is to enthuse a new generation of nerds. And like it or not, ARM and x64 is all they've ever known.

_________________
Doing stupid things for fun...

@Karlos

Quote:

Karlos wrote: If you want proper 64 bit and SMP, there's AROS and AxRT.

One fixes none of the issues while introducing some more while the other is little more than using Zune/Intuition as yet another GUI toolkit when running Linux.
Interesting, just not interesting enough atm.

As for PPC,NG or whatever being "viable" that is not the point as at the very moment you add an API for special "high" memory you have created another "NG" solution.

You know something that does require SW to adapted to use that extra functionality. Apps that then won't work for on non-NG Amigas (or other NG variants).

You are off course free to not go down that road and stay 68k/31Bit/Os3.1-3.x forever but if you do go down it, go down it all the way.

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

@Kronos

It sounds a bit rich, really. As if NG has somehow stepped out of the shadow of 3.x/68K and it's limitations.

Name one single feature that's not simply an incremental improvement over what was already possible with 3.x/68K and that *cannot* be achieved with 68K because of some architectural improvements that migration to PPC enabled.

I can't think of any, not here in 2024. Really there should be SMP and full 64 bit support now. Where is it? Sure there are improved API for almost everything, but they are we evolutionary, not revolutionary.

The only thing the PPC ever had going for over 68K at the start was speed and the promise of continuation when 68K was halted. At the start, that truly enabled things no physical 68K could match, it just wasn't practical.

Even that's no longer the case. PPC is just as dead as 68K was and there are now classic Amigas with PiStorm/Pi4 in them that are faster than a significant proportion of NG machines. And then there's UAE, which just scales with however fast the underlying x64 is.

There's nothing in the NG bag that couldn't be backported and run perfectly well on these configurations because they never stepped outside the original limitations anyway.

And I expect, they never will.

_________________
Doing stupid things for fun...

@Karlos

And yet you are the one suggesting that a virtual 68k solution with high memory support would not qualify as "NG".....

Last edited by Kronos on 22-Dec-2024 at 04:36 PM.

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

@Kronos

Firstly I'm not proposing that any kind of private 31 bit to physical 64 bit should be done, only that it could be. Or you could go true 64 bit because the CPU you are *natively* executing on is 64 bit. We all know why this isn't happening. None of the legacy software would work because the 32-bit pointers are encoded into structures everywhere. Only a complete recompile of everything would work.

By contrast the (real) 68K has no option to run 64 bit because no 64-bit variant of the CPU exists. However, you *could* make use of extra memory by using such a private mapping scheme when running in virtuo on a 64-bit bare metal emulation.

Does that make it "NG" ? No. Why? Because NG isn't 64 bit anyway. It just makes it a more capable current gen, which is all I'm interested in anyway.

_________________
Doing stupid things for fun...

@terminills

Quote:

Technically it's not judging but punishing and you are correct which is why I wouldn't.

Well, in a way it would be a judgement as well, viewing the PPC community as unworthy of your software. Which I think we wouldn't deserve - we're trying to preserve the Amiga spirit as much as anybody else around here or elsewhere.

Glad you're not taking the guy as a representative sample You know there has been a very good response to FinalWriter news at Amigans.net, and people are waiting for the OS4 version, cash in hand.

_________________
The Rear Window blog

AmigaOne X5000/020 @ 2GHz / 4GB RAM / Radeon RX 560 / ESI Juli@ / AmigaOS 4.1 Final Edition
SAM440ep-flex @ 667MHz / 1GB RAM / Radeon 9250 / AmigaOS 4.1 Final Edition

Ironically, what I'm proposing is basically the same as MorphOS. MorphOS has Quark, that in turn hosts the ABox. Everything people think of as MorphOs is just the ABox, a 32 bit PPC runtime that sits on top of a deeper substrate that runs on the actual hardware, which can be 64-bit.

A bare metal emulation / HAL that incorporates a high-performance JIT 68K would then host a similar "ABox", one that's 68K instead.

Imitation is the sincerest form of flattery, right?

_________________
Doing stupid things for fun...

@Karlos

Quote:

Does that make it "NG" ? No. Why? Because NG isn't 64 bit anyway.

So "NG" means nothing hence you can't call it "NG" when you add something to it.

In the end it's all the same, once we added stuff like CGX or AHI to Amigas we pretty much created NG Amigas. Or at least when we started using SW that relied on those APIs to be there.

You can call it what you want but once you write SW for faster than real life 68k that uses more RAM than stock pre 94 Amigas or needs APIs that are not supported by them you have already fractured the SW market.

Do it for a specific variant of fake retro HW and your a small splinter.

Or only do SW that would work with 30+ year old HW giving users of more recent HW little more than basic creature comforts.

But back to the topic:
Plenty faster/better/cheaper option to run old SW than whatever the OG post hints at and if you won't to run "new" SW it just can't be more than another compromised half measure in what is already a crowded field of fake_retro/EMU/NG options.

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

@Karlos

Quote:

Karlos wrote: Ironically, what I'm proposing is basically the same as MorphOS.

So you are proposing an 68k clone of a not_NG_by_your_definition PPC-OS and than add NG features to but just don't call it NG

Got it, makes perfect sense

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

@Kronos

I don't think you get it, but that's fine because you clearly aren't the target audience. You are taking a very simple proposition and trying it tie it in knots for god knows what reason.

Look, it's super simple. As an Amiga enthusiast, I want to run AmigaOS. I don't just want to run it on my classic Amiga, which will surely die someday. I also don't want to just run it in UAE as an application on a a different OS. It's fine for quick sessions on a go anywhere device, but what I really want there is something closer to a bare metal emulator on a desktop. On hardware that's readily available, upgradable and replaceable.

As a developer, I want to be able to write software and compile it exactly once, for one architecture that then runs on as many of these 68K platforms as once, including under emulation on NG systems.

What I don't especially care about, at all, is supporting PPC itself. I just see no value in it.

Also, for the avoidance of doubt, things like SMP and extended memory support are things that I *think* might be made simpler by a in virtuo implementation and thus potentially achievable. I'm not especially advocating for it. So don't get that conflated.

Last edited by Karlos on 22-Dec-2024 at 05:56 PM.

_________________
Doing stupid things for fun...

@OlafS25

Well, there is Vampire standalone also. A600 came out. Now they wish to bring this A1200 out.

Between the three of them, AFAIk the Vampire standalone can run Amiga OS on it. Thats a bonus for those making it duel boot, and or swapping CF / SD cards between AROS and Actual Amiga OS I would say.

Unsure what others there are, would one include all the Pies (Definitely emulated) and A500 mini?

While we are here should we add the C64 also mini was it?

We definitely spoilt atm also if you own retro hardware for accelerators and or expansions...

That still seems to leave the new generation at a loss then for now somewhat.

_________________
Courtesy of SAM440Flex & Amiga OS4.1 only
Flex is 800mhz
A1000 with Classic 520 Amiga OS3.2.1
AmiKit 12
MorphOS PowerBook G4 (which can play youtube vids)

https://blitterwolf.blogspot.com

@Hans

Interesting Point Of View, not necessarily accurate.

Example or Exibit A :
You create a youtube page, after a few years you still only have about 100 or 200 subscribers, yet your friend has 2000 or 20000 and yet they did the (lets say it was about music aye) same type of genre music, yet they have more subscribers than you.

Better yet, lets say they only had 2000, but had 50k views and any one of yours only had 100 - 300 views, this still doesnt represent an accurate amount of people that may be interested in an item, especially if they didn't know it ever existed.

Also algorithms play a part in this too, google search, etc,e tc, awareness.

Your friend created their channel same day lets say, but they also went out of their way, maybe posted elsewhere, or anything other site, or sub genre, etc, etc that they did, then they always ensured they linked their youtube music...

For arguments sake lets say you didnt.

Another interesting Point is, Amiga facebook pages, theres one with 68k users, another with like say 500, which one should we believe is more accurate of interested people in Amiga then? lol

ANYHOW

Long winded story that was, I apologies and completey understood your point and saw some degree of value in it.

BUT

In your newsletter case, I just noticed I'd have to sign up, I barely like signing up to begin with to anything new, as there are already lot of damn passwords to recall for other more pressing things in daily life.

Is there a way to bypass signup, and yet still view the newsletter for example?

What if you allowed people to visit your site and read the newsletter without signing up, would there be more traffic, more people reading?

Well, these are just open thoughts of possibilities

_________________
Courtesy of SAM440Flex & Amiga OS4.1 only
Flex is 800mhz
A1000 with Classic 520 Amiga OS3.2.1
AmiKit 12
MorphOS PowerBook G4 (which can play youtube vids)

https://blitterwolf.blogspot.com

@agami

But I can use AmiKit and Emulate all the 68k programs just fine, this is where I would tend to almost agree with the other chap crying Wolf.

Others using Amiga Forever.....

Others using Win UAE......

Wasnt there another not long ago to emulate OS4 also?

Oh yeah then theres FlowerPot....

Did I forget PiAmiga? (BTW Linux software package has this too, amazing)

You see we have more than ample ways to have Amiga OS in Emulation form

Maybe I misread your post?

Are you suggesting more the ease of use for a newcomer to simply plug and play, a bit like the A500 Mini was?

_________________
Courtesy of SAM440Flex & Amiga OS4.1 only
Flex is 800mhz
A1000 with Classic 520 Amiga OS3.2.1
AmiKit 12
MorphOS PowerBook G4 (which can play youtube vids)

https://blitterwolf.blogspot.com

terminills Quote:

No one using an Amiga or Clone really cares. PPC is dead, technically 68K is dead even though it's quite active for Amiga standards. but at the end of the day most just want to use the programs they want and enjoy themselves.

The virtual 68k Amiga pseudo standard has moved forward in two ways.

1. Performance

2. AmigaOS 3 updates

Some people continue to say performance does not matter as the old 68k software does not need it but it is the primary reason for what some people are calling 68k+. Without performance, the 68+ would actually be 68k- as less of the 68k ISA is used, especially any MMU. The ISA is not enhanced but actually moving backwards as the AmigaOS 3 updates are only compiled for the 68000 which does not include direct FPU use or even full 32-bit support (e.g. 32x32=32/64). The 68000 ISA standard is from the 1979 16-bit 68000 CPU lacking pipelining, caches, a 32-bit ALU and a 32-bit data bus (the AmigaOS for the 68000 is 32-bit because the ISA and pointers are 32-bit). Other architectures are moving forward while the 68k technology and support is moving backwards. This may be acceptable to people that just want to run 68k Amiga software faster than back in the Commodore era but they sacrifice 68k technology for foreign technology that is supported instead. It is very much performance driven while there are no ISA enhancements, 64-bit support, SIMD support, SMP support, etc. Most compiler and emulator developers will not support ISA enhancements for a virtual 68k Amiga either. This is not a way forward even though the push is driven by the desire for increased performance, emulation is an inefficient way to deliver it and modern 68k enhancements are not possible.

Karlos Quote:

Not necessarily. Not all code is amenable to being fully translated ahead of time and I suspect that under AmigaOS/68K the corpus of code that would actually work well here is small. Which then defeats the whole point of the endeavour. Of course, you can have both mechanisms. However, in fretting over JIT you are consistently overlooking an important benefit: JIT compilation and hot path optimisation can produce native code that's fine tuned for your exact configuration at the time you run it, taking into consideration many factors that no AOT compilation can reliably try to predict.

AOT translation of code ahead of time has the advantages of being able to perform some static data endian conversion and being able to perform more optimization than JIT where we see ARM performance suffering from lack of instruction scheduling. It may be possible to better optimize the JIT hot spots but trying to optimize code in heavy use is undesirable. Trace driven optimizations can improve JIT performance but increase startup latency and jitter while also using more memory. There are not many virtual systems anymore like JAVA virtual machine (JVM) but AOT compilation is used by the Android Runtime (ART) replacing Dalvik JIT.

https://en.wikipedia.org/wiki/Android_Runtime#Overview Quote:

Unlike Dalvik, ART introduces the use of ahead-of-time (AOT) compilation by compiling entire applications into native machine code upon their installation. By eliminating Dalvik's interpretation and trace-based JIT compilation, ART improves the overall execution efficiency and reduces power consumption, which results in improved battery autonomy on mobile devices. At the same time, ART brings faster execution of applications, improved memory allocation and garbage collection (GC) mechanisms, new applications debugging features, and more accurate high-level profiling of applications.

GPU intermediate languages for unified shaders also switched from JIT to precompiled AOT like compilation. Javascript JIT is still used but there are alternatives. JIT is losing popularity due to lack of efficiency/competitiveness and whole virtual machines are practically dead. If performance and value are the goals, native code for real hardware can not be beat.

Karlos Quote:

Emu68 and UAE both utterly shatter the performance of all extant 68K silicon (your ASIC is not real and likely never will be) and for something intended to be backward compatible with software that runs on extant silicon at tens of MHz, you are worried about the performance overhead of said emulation? Just go back and look at the Lightwave tests. Some of the most arithmetically dense, unpredictably branchy, scattered memory access workload imaginable (that's raytracing). UAE and Emu68 both ate it for breakfast and asked for more. Honestly, I don't get the objection. Sure I get that for some people only physical silicon is real and they'll never settle for anything less. That's fine, there are plenty of accelerator solutions available today. Nobody is taking anything from them by providing an alternative that can run on commodity hardware.

Physical silicon is what is "real" not only for users but for developers. Sure, the old 68k silicon is ancient but the 68k Amiga can be updated, enhanced and modernized which has already been demonstrated by FPGA 68k Amiga hardware. There are many 68k CPU cores, Amiga chipset cores, GPU cores and modern SoC IP available to license and as open cores. Real hardware is relatively risky, hard and expensive but it would give the 68k Amiga a chance to compete again. Virtual machines are easy but a dead end end of life choice that lacks value. If Amiga hardware is wanted, it needs to offer competitive value to be mass production successful and there is only one choice which is 68k Amiga hardware using more modern silicon.

@agami

They'd need to make the AROS installs easier for non tinkeres.. Especially their internet side of things.

This assuming that it doesnt come pre installed.

Having said that, if its similar tot he Vampire Apollo os install....

Lets just say the other versions, not sure who updates them, but if you download Aspire, AROS x86, or what not, those installs can take a full day to iron out if it doesnt wish to play ball.

Infact, the trusted netbook that always had accepted AROS in the past, well, AROS no longer installs onto it, but it easy loads into the OS via Disc, or usb key.

Anyway, just saying, if the community wants more Amiga users, needs ot be far less technical to install, make it more like Linux Mint, Debian, Zorin OS, damn, just pick one, all easy installs even for someone who doesnt know much, internet access also easy on said Linux distros...

Peppermint OS, Etc, etc, there's many.

_________________
Courtesy of SAM440Flex & Amiga OS4.1 only
Flex is 800mhz
A1000 with Classic 520 Amiga OS3.2.1
AmiKit 12
MorphOS PowerBook G4 (which can play youtube vids)

https://blitterwolf.blogspot.com