If you had the chance to build a new Amiga

[rcman]

Hi All,

I know with advanced processing power like the vampire and Amiga emulation working well I had a thought. If you could design a modern Amiga that had custom chips and enough power what would it be?

As an example I know I would design a motherboard that had custom chips as the Amiga has now. They would be a lot more powerful than what’s being used now.

Graphics, sound and blitter chips with serious advancement in their capabilities.
But maybe I’m going about this all wrong, I’m not a very good programmer or engineer. With the 8-bit guy designing his new dream 8-bit system. What would be your ultimate Amiga System?

Thanks
RC

[chip]

Interesting question, but without technical skills it's nearly impossible to answer IMHO

I would like a sound chip that can produce only one kind of music, so that music can be easily preserved

I mean, no more exotic formats (which i love anyway), but only one music format

[Tigerskunk]

I think modern and Amiga is a paradoxon.
There is enough other stuff that is "modern".

[deimos]

We need to roll back the clock and get a QNX style real time micro kernel, on a processor built from scratch to ensure that process isolation is enforced all the way down and even further, if there's a virtualisation layer to consider. Designing a system for modern needs is probably impossible, but I'd take a performance hit to make things safer.

[Jope]

- A3000/A4000-esque architecture
- still 68k based, fast 060 or apollo team 080
- the fast side of the motherboard would have to be 060 or 080 local
- modern hw like usb, dma sata, dma ethernet on the fast side
- the slow side of the motherboard should have old style custom chips that are switchable between ocs, ecs, aga, and modern modes
- all screenmodes available via hdmi, modern screens seamlessly switchable/draggable among with classic screens for maximum compatibility with old os legal stuff. Old screens perhaps pixel doubled or tripled so that they aren't postage stamp sized when you drag down a fullhd screen..
- expansion slots PCI or PCIe, better forget about Zorro III..

Pie in the sky dreams of a mostly backwards compatible but expanded Amiga. :-)

[Samurai_Crow]

One OS feature that I would have liked is the ability for macros could be embedded in libraries. AmigaDE could do it but not AmigaOS in any version. That way the huge split between system friendly code and hardware banging could be erased.

Actually, a bytecode like AmigaDE could have been helpful in running on different CPUs with only a minimal conversion penalty.

[-Acid-]

If i was designing an Amiga now I would scrap any idea of custom chips and use off the shelf PC parts. Custom chips are what finished the Amiga's competitive edge in the end once hardware manufacturers caught up overtook and left the Amiga behind in the dust. It would be a waste of money making custom chips now when you can just use existing processors and gpu's etc that other companies have already spent years and £millions developing, there is no way anyone could start from scratch now and design something to outperform today's hardware.

If a new Amiga was to become a reality it should simply be a better more streamlined and efficient operating system than what is available now similar to Linux using all standard hardware. Of course it should be designed with tinkering in mind like the Amiga was with scripts mostly in plain English and creativity tools supplied with the OS etc. For backwards compatibility I would hire Toni Wilen to make UAE a part of the OS built in somehow maybe even running seamlessly?

[grond]

My new Amiga would have a GHz 68k processor pretty much like the 080 but with an SSE type coprocessor unit on top. It would look like a computer keyboard but the cable coming out its back would be the video output because the whole computer would be inside the keyboard. It would have an SSD, a couple of USB ports and an SD-card slot. Computing-wise it would only be in the tablet computer category. The OS would mostly keep the old API but get memory protection, multiprocessor support, multiuser support and ownership concepts. Old programs would run within a single context legacy OS task that could be terminated and reinitialised without having to reboot the system. Anything that bangs the hardware from the legacy system would be intercepted and emulated. In this way most of the legacy software stuff would run seemlessly inside the new Amiga. Incompatible stuff would have to run in UAE.

[Valken]

Have to echo Acid's comments.

Modern x86-64bit CPU with AROS with OpenGL, OpenCL, OpenAL type HW extensions, and bare metal UAE 68K, PPC emulator baked in. A lite emulator "wrapper" app with plugins to create virtual HW or chipsets on top of UAE 68K. This would allow many groups of emulator creators to pool resources for CPU, audio, video, file system and etc.. access.

Hopefully AROS with AmigaOS extensions exposing all the new "hardware" features would give incentives for people to make Amiga like games with more power to them.

Here is an example of a game style that should be achieved on the Amiga Next:

https://store.steampowered.com/app/1...es_Your_Grace/

[ Show youtube player ]

Beautiful Pixel artwork, modern 3D lighting, shadows, reflections, ambient effects, good voice, sound and music. Basically a ton of polish and love went into this and feels like a Cinemaware game.

Imagine our classic Amiga games like this and new sequals and remakes?!

Technically this would run on PC, Mac, Xbox, Playstation and any x86 based HW today but still has that Amiga feel or heck, any classic 68K platform feel, and still run AmigaOS, Atari TOS or Mint, Classic MacOS and etc...

I personally REALLY like the concept of AROS HOSTED on top an OS. Its just that I need to install Linux to make the best of it as Windows Hosted is not fully supported yet - audio, network and 3D.

Imagine near native access to Amiga, Atari ST, Mac, Sharp X68000, FM Towns, PC98, 3DO and all console type at native hosted PC HW speeds in a light wrapper acting as native apps.

[meynaf]

I would create a new cpu, similar to 68k but different, with the same original design goals that made the 68k but taking programming experience into account so it is even more a beauty to code in asm.
The hardware would be designed to have easy to use interfaces that would abstract underlying implementation.
The OS would be rewritten from scratch, with both a low level api system similar to the actual one, and a more high level system on top of it. It would be highly configurable, making memory protection, multi-user, and whatever, all facultative.
Compatibility would need emulation but with a multi-ghz cpu and powerful enough gpu it's not a big deal.
Rather than really a new Amiga, it would be the Amiga's offspring.

[Glen M]

A modern Amiga would not be in custom hardware but rather in the OS i.e. the same as Windows or mac os, they run on 64k hardware. If you want a modern Amiga someone needs to port OS4 to x64 architecture. You can sort of get there with ppc hardware but even that now is 15 year old. AROS works on some x86 hardware but again it's old. I understand there is an x64 port of MORPHOS in the works. If that does happen with support for say a Ryzen CPU and an rx GPU then I think that'll be as close as you'll ever get.

[Old_Bob]

Interesting question. If I allow my imagination to run with the ball a little...

I think I'd like to see a modern multicore 68060 CPU with, let's say, 16 cores running at 4500mhz each. Along with maybe 32mb of cache memory.

On top of that, it'll obviously need some kind of modern, fully featured 3D graphics hardware that is somehow still compatible with OCS/ECS and AGA (!!!). Likewise we'll be needing a new sound system capable of maybe 16 channels at a 96khz sampling rate and 24bit resolution. Chuck in an off-the-shelf network solution from somewhere and we're good to go.

If someone wants to lend me £100 million, I'll make it happen.

B

[Hewitson]

You would need a hell of a lot more than 100 million to design that sort of CPU. Intel or AMD probably wouldn't even be capable of it.

[d4rk3lf]

Originally Posted by Old_Bob

On top of that, it'll obviously need some kind of modern, fully featured 3D graphics hardware that is somehow still compatible with OCS/ECS and AGA (!!!).

As Meynaf pointed out, when you have that of a jump in hardware, you don't really need hardware compatibility with previous chipset, because with software emulation it will eat it for breakfast.

I agree with Hewitson that 100 million is too small.
Maybe couple of billions, and few years of development.... but even then...
(I am waiting patiently that some Amifa fan becomes multi-billionaire )

This brings me to the topic question:
I doubt, even if you had almost unlimited money, you could create something really ground-shaking to the industry.
Except... completely abandon silicon, and create chipset with material that is better then silicon (someone more knowledgeable then me, could probably tell more details on this)
Now, that's the new Amiga I'd like to see.

[Old_Bob]

Originally Posted by d4rk3lf

I agree with Hewitson that 100 million is too small.
Maybe couple of billions, and few years of development.... but even then...
(I am waiting patiently that some Amifa fan becomes multi-billionaire )

A billion, or two? Okay, then. If everybody on the planet contributes £1, we'll have plenty.

B

[Amigajay]

I would take a conservative approach and go for something in-between a Spectrum Next and Pico-8 in terms of forcing limitations on the game designer to both make it easier for a level playing ground.

Sprite capabilities in excess of Neo-Geo, with a max resolution of 320x256 for games, along with maximum number of colours 256 (the height of pixel art imo) and advanced blitter, copper and audio within set limits.

Imo its pointless aiming a new Amiga with modern specs, the software support would never be there with the budgets needed to make a modern looking game, and you might as well get a cheap PC for 1/3 the price. (Kinda ironic tbh!)

[d4rk3lf]

Originally Posted by Old_Bob

A billion, or two? Okay, then. If everybody on the planet contributes £1, we'll have plenty.

B

It's maybe easier to convince every person on the planet to donate 1 dollar, then to find 2 Amiga fans that could agree how to spend this 7-8 billions for Amiga future.

[skan]

Originally Posted by Steril707

I think modern and Amiga is a paradoxon.
There is enough other stuff that is "modern".

Originally Posted by Amigajay

I would take a conservative approach and go for something in-between a Spectrum Next and Pico-8 in terms of forcing limitations on the game designer to both make it easier for a level playing ground.

Sprite capabilities in excess of Neo-Geo, with a max resolution of 320x256 for games, along with maximum number of colours 256 (the height of pixel art imo) and advanced blitter, copper and audio within set limits.

Imo its pointless aiming a new Amiga with modern specs, the software support would never be there with the budgets needed to make a modern looking game, and you might as well get a cheap PC for 1/3 the price. (Kinda ironic tbh!)

My thoughts exactly.

[rcman]

There's many options. I would not make it with anything X86 based that's for sure. I would like you suggested a Motorola derivative.

I know Vampire made "Amiga in a Box" basically made the an AGA machine running pretty fast but it would need to go way beyond that.

I think first for foremost it would need to be a programming machine. That could accomplish anything with graphics. sound to the most extreme.

RC

[AmigaHope]

If by "modern" you mean truly up to date, I wouldn't include any of the Amiga-style custom chips. There's simply no use for them. The stuff that they were good for has been completely subsumed by the programmable vector processors we call GPUs. They cover the concept of the old Amiga copper and amplify it by a zillion. There is no longer any need for dedicated 2D hardware sprites, playfields, etc. because custom GPU hardware can do it all fast through simple code.

Custom chips I *would* include -- an AI-specific ASIC, this is where the future is going. I would also include a big FPGA (and appropriate support in the OS to handle the low rate of context switching that would be needed to multitask with it -- this is new ground). Bits of the fabric would have to be allocatable in some way, etc. and how to share it between processes.

I admit the FPGA is a bit of a stretch, but there are cases where it's a win (complex non-parallel function that takes many cycles on a general purpose processor that can be rendered into a single-cycle operation in an FPGA).