Next gen Amiga

[jediknight]

the nextgen amiga should be a new 68080 a4000 with loads of memory and improved vram maybe use a pci card ill buy two.

[Locutus]

Originally Posted by meynaf

Transmitting some info to some OS for the GPU to use, isn't what I'd call "directly".

Theres register level documentation, and you can perfectly well write directly against it..... if you really insist you can do bare hardware graphics without on your own code.

[indigolemon]

Originally Posted by jediknight

the nextgen amiga should be a new 68080 a4000 with loads of memory and improved vram maybe use a pci card ill buy two.

I too enjoy off topic statements with no basis in reality.

[saimon69]

Originally Posted by TurboCrash

I hate saying this because I love 2D games and loathed the first fully textured 3D games, but the groundbreaking-one-step-ahead that characterized the Amiga was taken by the Sony Playstation and by that time the PC was already doing pretty much everything the Amiga could do and some more (that was groundbreaking in the later half of the 80's). So in this analogy the Amiga would have had to release a new model with a 3D accelerator card before the competition to be as groundbreaking as the Amiga 1000 and 500 were before. And then a *nix based system and a smart phone, etc...

AFAIR that was the plan for AAA, but being that a GOOD thing, the C= management killed it

[meynaf]

Originally Posted by Locutus

Theres register level documentation, and you can perfectly well write directly against it..... if you really insist you can do bare hardware graphics without on your own code.

But current OSes won't let you do that, will they ?

[Dunny]

People are hitting the hardware directly on the Pi these days, including the 3D Acceleration. It does require that you don't run a traditional OS though, you boot directly into the application.

[Megol]

Originally Posted by meynaf

Transmitting some info to some OS for the GPU to use, isn't what I'd call "directly".

Neither would I. But that wasn't what I was referring to: directly controlling the GPU by writing registers and memory fields according to the documentation.

This would probably be easiest in something like AROS or one of the assembly OS available (MenuetOS, KolibriOS, BareMetal OS, ...).

Because they didn't know that a lot of what made it a great machine wouldn't have been possible without this.

Possibly. Or perhaps they realized that abstraction layers were a requirement for future expansion.

Remember that the Commodore NG Amiga was a very different beast than the so called NG efforts existing today.

Don't get me wrong: direct hardware hacking is fun but it's very hard to have continuous hardware improvements with it unless one does something like early Playstations or the PC VGA standard: keep old hardware support as a part of the new system.

[idrougge]

Originally Posted by Megol

But is this only positive? Before throwing things at me remember that Amiga OS while advanced for its time have severe limitations even in the NG forms.

Old programs already have to run under emulation so why not continue this for the OS too? A modern OS strongly inspired by the original with subsystem support for the legacy OS format could be nice IMHO.

The difference is that the next-gen Mac switched to an entirely different OS. There is absolutely nothing in common between MacOS 9 and MacOS X, and not just because of technical reasons.

Windows, OTOH, has stayed the same system since it first came out, doing small incremental updates year after year.

[Dunny]

Say what you will about Microsoft, but their backwards compatibility in Windows is quite a feat.

[Bruce Abbott]

Originally Posted by Megol

Of course you can code for it directly if you choose the right GPU. Intel and AMD have released good information in the past at least. But why?

Why indeed. You spend countless hours struggling to produce code that only works on one particular hardware configuration, then upgrade your pc and all that effort is wasted. The Amiga's main advantage over a PC was standardized hardware. Even back in the days of DOS and VGA, compatibility issues on PC's were rife. Then manufacturers started making cards with secret hardware that only worked with their own special drivers, and direct programming died. And compatibility is still a constant problem on PCs.

Not that direct programming was considered a central part of the Amiga concept by the owner(s).

99% of Amiga games wouldn't exist without direct programming, and who do you think made them? Many of us started out on home computers such as the ZX Spectrum, Commodore 64 or Amstrad CPC, where direct programming was essential. Without that scene I doubt the Amiga would have gone anywhere.

Imagine if Amigas could only be programmed through system libraries and the hardware details were a closely guarded secret. Apple did that with the Macintosh and its gaming reputation suffered. Luckily Commodore didn't. Instead they documented the hardware and showed us how to use it directly, through the system, or both. The first code I ever wrote for the Amiga was a magazine type-in that accessed the hardware directly. It took me nearly a year to fully understand system programming, but many other coders never bothered because direct programming was easier and more rewarding.

I also tried direct programming on my first PC, an IBM JX. Just getting a serial mouse to work was a nightmare, and the graphics code I wrote would not work on other PCs. I learned a lot about PCs on this machine, but it also showed me what a mess that architecture was (and it never got better).

[Bruce Abbott]

Originally Posted by Dunny

Say what you will about Microsoft, but their backwards compatibility in Windows is quite a feat.

But in practice Windows compatibility is spotty - and not just for games. I am currently working on drones for two companies and both are having serious compatibility issues with Windows 10. The software has so many layers and bolt-ons that figuring out how to fix it would take months, and even if we do there's no guarantee it won't break again on the next system update. And since updates are required for security, the nightmare will continue!

In comparison, programming the Amiga is a breeze. Just stick to plain 68000, 1.3 library functions for OCS chipset or 3.0 for AGA, and your code will work on practically any real Amiga (maybe not on 'NG' machines, but they are not true Amigas). Debugging the Amiga is easy because there's no memory protection so you can poke around anywhere without fear, programs are small enough to disassemble and step through (and 68000 machine code is much easier to understand than Intel), and if the system goes unstable you can reboot in seconds.

Trading compatibility for performance is a non-starter for me. PCs suffer because of it. The Amiga doesn't have to. When Commodore was talking about going Power PC I thought it was a mistake. I suggested that if they insisted on abandoning the 68000 they might as well just go Intel. Perhaps if they had we might have gotten newer more powerful 'Amigas', but current trends in the 'NG' Amiga scene suggest that it would have fractured the market. We are lucky that Commodore died before they got a chance to mess it up.

That's why I am excited about Apollo's Vampire accelerators, which speed up real Amigas to NG performance levels while still maintaining virtually 100% compatibility with standard 68000 code. A few years ago I looked into what was happening on the Amiga scene and it seemed to be dead, despite the availability of (ridiculously expensive) NG hardware. Now there's a resurgence of interest in classic Amigas, and people are (re)discovering their appeal. Before now I would not have believed there were enough buyers for over 2700 Amiga accelerator cards, nor that I would ever consider spending hundreds of dollars on one. Let's hope this trend continues!

[MigaTech]

Originally Posted by Bruce Abbott

In comparison, programming the Amiga is a breeze. Just stick to plain 68000, 1.3 library functions for OCS chipset or 3.0 for AGA, and your code will work on practically any real Amiga (maybe not on 'NG' machines, but they are not true Amigas).

Is this completely true? What about the 2nd Gen chipset ECS, differs from OCS 1st Gen, hence the heavy compatibility issues of the early 1990's.

[Daedalus]

It all depends on the level at which you're programming the Amiga. If you're banging the hardware, doing direct jumps into ROM code and using absolute addresses for RAM, it will be sure to fall over as soon as you try it on another configuration.

At the other end of the scale, if your program uses the OS and doesn't bang the hardware at all, it should work on every machine from that version on, including NG systems. Coding in this way would also let your game take advantage of things like graphics and sound cards, USB input devices and so on, even on a classic machine or a Vampire.

That's why I am excited about Apollo's Vampire accelerators, which speed up real Amigas to NG performance levels

Really? Current Vampire setups can outperform a high-end 060 system in certain cases, but they're still a long, long way off even an A1-XE, which is pushing 15 years old at this stage. And if you bring MorphOS into the equation, you can find yourself a 1.8GHz NG machine for less than the cost of a Vampire board.

while still maintaining virtually 100% compatibility with standard 68000 code

68000 compatibility isn't the issue with NG systems, it's banging the rest of the machine's hardware that causes problems.

[Dunny]

I suppose that any NG AmigaOS needs to be able to run classic binaries and media images regardless of the host architecture. I believe some modules of AROS attempt to do this, but as an OS AROS leaves a lot to be desired as of yet.

And it needs to do it flawlessly too - the user should never be able to tell that it's happening.

[jPV]

Originally Posted by Daedalus

And if you bring MorphOS into the equation, you can find yourself a 1.8GHz NG machine for less than the cost of a Vampire board

Or even 2.7GHz machine. PPC Mac prices have really dropped in last couple years and soon anything over 100e is too much, even of the fastest models

[Daedalus]

Indeed, even faster I was just thinking of a 1.8GHz machine I had myself and gave away for free. Only later when support for it was introduced by MorphOS did I regret it

[meynaf]

Originally Posted by Daedalus

At the other end of the scale, if your program uses the OS and doesn't bang the hardware at all, it should work on every machine from that version on, including NG systems. Coding in this way would also let your game take advantage of things like graphics and sound cards, USB input devices and so on, even on a classic machine or a Vampire.

You won't get far with not banging the hardware at all, at least on low end machines.
If you want to have decent performance you'd better access your screen's bitplanes directly.
If you want to play Protracker then you *have* to access Paula directly - as audio.device has dma wait problems and simply can't do this.

[OlafSch]

to make a short comment myself too...

IF Commodore would have survived it would have dropped most of the chipset in favor of standard components. It is simply a matter of resources and money, no chance to compete here. I myself owned a A4000 (68030) back in the days with graphic card added (Picasso 2). It was a huge difference if you are interested in f.e. productivity software. The chipset was good for gaming but not for serious software. It was still amiga for me . But to compare A4000 with added graphic card to what people now call NG hardware and say no difference is not true in my view, it still was Amiga with 68030, there were still the same expansion slots, it runs the same OS and the original chipset was still there even if not used depending on the software. Amiga always was (for me) mainly hardware, at that time hardware was way more important than the OS. Of course that changed over the years, hardware has become cheap and widely available and now OS (and software) makes the difference. But even if I see it this way I see Amiga as a retro device so Vampire at least comes next to what you could call "Amiga NG". MorphOS running on used Macs is still MorphOS, Aros running on X86 the same and X5000 is basically a standard board with a different processor and uncompetitive price but not Amiga. But that is a personal view of course

[Bruce Abbott]

Originally Posted by MigaTech

Is this completely true? What about the 2nd Gen chipset ECS, differs from OCS 1st Gen, hence the heavy compatibility issues of the early 1990's.

I don't remember 'heavy' OCS/ECS compatibility issues. I do remember games that wouldn't work on Kickstart 1.3 or 2.x, 1MB RAM, or even a second disk drive. These early incompatibilities were mostly caused by crap programmers who refused to follow the rules or were just incompetent.

Any incompatibilities between graphics chips etc. on the Amiga are not due to the documented hardware, which is fully compatible if used correctly. This is quite different to PCs, where compatibility is maintained via custom drivers which interface to the constantly changing incompatible hardware underneath. On a Windows PC it is virtually impossible to access the hardware directly, and even if you could the documentation is poor to nonexistent. That is by design, and is one reason why in its heyday the Amiga beat the pants off comparable PCs.

[Bruce Abbott]

Originally Posted by Daedalus

68000 compatibility isn't the issue with NG systems, it's banging the rest of the machine's hardware that causes problems.

I had an A3000 with 68060, so I know that the CPU can be a factor. But of course the hardware is also a factor, which is why I want an accelerator card (unless someone can produce a chipset which is compatible with classic Amigas).

Current Vampire setups can outperform a high-end 060 system in certain cases, but they're still a long, long way off even an A1-XE, which is pushing 15 years old at this stage.

I'm not familiar with all variants of the AmigaOne, but they seem to just be overpriced under-powered PCs that run 68000 code via emulation, right? How fast do they do it? Do I really want a big box with a PC motherboard in it? (no).

And if you bring MorphOS into the equation, you can find yourself a 1.8GHz NG machine for less than the cost of a Vampire board.

MorphOS isn't Amiga OS - and a real Amiga is more than just the OS. If going NG means having an 'Amiga' that isn't really an Amiga then I am not interested.

Clock speeds are just a numbers game. My current PC has a dual core CPU that runs at 2.8GHz, but apparently that isn't fast enough today (I have been reliably informed that if I want to develop programs for a 2kB PIC MCU I need an I7 with 16GB and a SSD). And since I refuse to 'upgrade' from Windows Xp my PC is now a security risk and I have been branded a 'criminal' for using it on the Internet.

Modern PCs have a ridiculous amount of power which is largely wasted, still have lots of issues and are getting worse. Programming them is no fun either due to the huge amount of crap you have to deal with. Unfortunately it seems that 'NG' Amiga is going the same way.