The Future of DTP, Illustration and general GFX/Audio using RISC OS desktop computers (Is 64-bit!)

With nearly 100 replies to this thread, I’d like to make a summary of what’s written that is relevant to the topic – making a 64-bit version of RISC OS written in C able to run on multicore ARM CPUs, latest version (ArmV9/Neoverse) (to run on very very fast GFX/audio creating/editing workstations). With this new version Arm has “cut the umbilical” to earlier versions by introducing brand new instruction sets and ditched direct 32-bit support.

This implies a total rewrite of the OS with the aim of maintaining look and feel – “the RISC OS experience” in full, 100% – with the core and modules reflecting new realities regarding the new 64-bit V9 instruction set, able to address arbitrary amounts of RAM, terabyte hard disks/no upper limit to files per directory, support latest versions of USB/HDMI/PCIe, etc.

It would appear that the current companies developing 32-bit RISC OS are unable, primarily because of economic considerations, to take on this rather extensive project even if 64-bit RISC OS (the core) written in C can be applied to any 64-bit ARM device of which there are billions – “just” add custom GUI.

Without a big company/investor footing the bill here, taking an interest in RISC OS’ survival, (for instance Arm Ltd., worth some $40 billion or so) it will become more and more irrelevant as a viable computer platform compared to others until it disappears some years hence. It’s already on that track just by being 32-bit. Ok for hobbyists, decreasing interest from professional users. *It truly deserves better.

• It would appear “the old guard”, those who were fissioned off from Acorn, are no longer a factor in Arm Ltd’s management- and engineering/R&D-staff so the chances here are probably super slim unless they want a world-class operating system to go with their world-class processors, all “Made in U.K”. And, to be honest.. what great karma it’d be.

RISC OS – the world-class operating system. There is coffee on the screen, really.

What makes RISC OS more than just a BBC Micro on steroids is its user interface. Why then suggest having a custom UI on ‘devices’? “If I was going to go there, I wouldn’t start from here.”

ARM do not give a toss about RISC OS.

Who’s going to write your amazing graphics / video / audio editors? Just buy a Mac.

ARM do not give a toss about RISC OS.

That was already true in 1998 when I first started work at ARM. I moved from running an academic publishing office on Impression Publisher on RISCOS to documenting ARM assembly language on FrameMaker on a PC.

RISC OS – the world-class operating system.

Arguably true in about 1990.

Just buy a Mac.

Yup. Done that.

AArch64 has no STM or LDM instructions so a list of LDR/STR and and ADD/SUB are going to be needed

It’s easily sorted out using a set of LDP/STP, which keeps the stack aligned. If you have an odd number of registers to be preserved, simply add a dummy register to one LDP/STP. Oh, sorry, the compiler does that for you ;-)

all “Made in U.K”

I trust you understand it’s not really “British” any more? After the referendum tanked the pound (which has pretty much been heading south since then), it was an interesting investment for foreign money. Who didn’t really understand what they were buying and want to sell it on (but to Nvidea caused consternation).
The government attempted to interfere and block to keep ARM as a British concern, but that’s only about six years too late having already crowed about how sales like that meant Britain was, and I quote, “open for business”.

ARM might be based in the UK, but it isn’t British, not when the money and influence is coming from (currently) Japan’s SoftBank.

As for world class. Dude, I need to go lie down for a while.

Hey, as to “world class OS” I meant potentially, with the right funding and the right development team(s). The Arm CPU/GPU SoC’s already in existence (with MediaTek’s Dimensity 9200 in a multi-SoC workstation system being the carrot) will be a match to any other platform at that price level, provided a new SDK is made (yeah..) that is programmer friendliness incarnate so that software packages can be written to take advantage of this rabid performance potential, magnitudes above anything single-core/32-bit. And the speed of a 2-3GHz system makes the difference between programs written in (non-portable) assembler versus (portable) programs written in C quite insignificant. Plus ensure increased longevity for the operating system that would be RISC OS for the 21st century. It’s either that or dodo.

Why not get a Mac or M$ machine? https://www.theguardian.com/world/2013/jun/06/us-tech-giants-nsa-data – PS – It’s not about what you have to hide, but them.

> ARM might be based in the UK, but it isn’t British, not when the money
> and influence is coming from (currently) Japan’s SoftBank.

Yes, the Acorn RISC Machine is said to do really well. Wonder why its software/OS isn’t?

You seriously expect a smartphone chip to outperform a M2?

The ARM architecture has a number of licencees; RISC OS somewhat less so.

Why have RISC OS running on a phone when you can have a phone OS running on a Mac? :→

> You seriously expect a smartphone chip to outperform a M2?

As I wrote, in a multi-SoC setup it would. RO64 can be written to support that. And that “smartphone chip” can run in circles around anything 32-bit all by itself, never mind 4-6 of them working together. An 8-core Cortex X3 CPU is nothing to sneeze at! And the Immortalis G715 GPU.. well it is a bit more capable than good old VIDC.

> The ARM architecture has a number of licencees; RISC OS somewhat less so.

Would https://github.com/ARM-software help? Didn’t know one needed a licence to use Arm CPUs/SoCs. Does RPI pay royalties to Arm for use of their products? Does R-Comp?

Didn’t know one needed a licence to use Arm CPUs/SoCs

One needs a licence to manufacture (and sell) chips with ARM designs. One doesn’t need a licence to write or use software running on systems incorporating those chips, although in some cases other parts of the system that you need to know details of in order to write such software might not be publicly documented.

> You seriously expect a smartphone chip to outperform a M2?
As I wrote, in a multi-SoC setup it would. RO64 can be written to support that.

This is just getting silly now. We can’t even work out how to support pre-emptive multi-tasking without breaking every existing RISC OS application, you are now talking about not just multi-core, but supporting multiple socket machines – every current RISC OS application is single threaded and would not benefit.

If want to see what some multi-Soc machine can do, you need run a OS and applications which are designed from the outset to be be able to use it, and RISC OS doesn’t fit that description.

This is just getting silly now.

Only just now??

If hypervisored and sandboxed to the N’th degree, running on an emulated StrongARM or such, would that be an issue – software could be blissfully unaware of anything multi-. The silliness lies in letting a very promising operating system (primarily its GUI/WIMP system) go to the dogs.

I think this fellow is pretty clued-in; https://www.youtube.com/watch?v=jlRAnO1GR0U which says what must be said on this techno-evolutionary topic. Adapt or dodo.

> If want to see what some multi-Soc machine can do, you need run a OS
> and applications which are designed from the outset to be be able
> to use it, and RISC OS doesn’t fit that description.

Back in the day some 37 (!) years ago, some very clever people at Acorn decided to design a RISC CPU, architecture and instruction set. The very same (really) very clever people decided to write the OS in assembler in order to maximize performance of a 8MHz ARM2 with back then most impressive results, but no regard for portability, even to new (then) Acorn RISC Machine architectures which (now) Arm Ltd. has committed with its ArmV9 instruction sets currently puttering along on CPUs in the 2-3GHz range. So we can safely bet on portability this time around since the performance decrease is negligible in this league – yes? Even emulated for the time being.

Shame if “money” is the only hindrance regarding doing a RISC OS 64 version, close to 100% rewritten in C. It could shine brilliantly on you-know-what. Potentially. Be a bit HPC. Like the Archie was way back when.

Erk.

The huge difference is the scale of what’s needed. A few very clever people was all it took to design the original ARM CPU, and the original RISCOS. The world has moved on, and to excel in today’s world things need to be a lot more complicated, and a few very clever people isn’t enough.

Back in the day some 37 (!) years ago, some very clever people at Acorn decided to design a RISC CPU, architecture and instruction set. The very same (really) very clever people decided to write the OS in assembler in order to maximize performance of a 8MHz ARM2 with back then most impressive results,

But they did not manage to make it compatible with BBC MOS and its applications, as RISC OS 64 will not be compatible with RISC OS and its applications. If you consider this as true, your OS already exists: Linux + ROX + RPCEmu. Work done (but you can’t call that RISC OS).

So, revive the ROX project, help to make a better RISC OS 5 emulator, and make a distro with all of this. IMHO, it’s the easier path.

Nota: ARM 32bit is not dead. It’s for high perf market, not for embedded.
The ARM 32bit world is huge, and – IMHO – it would be easier to find ways to expand inside it.

For example, it would be interesting to have an OS made for some Cortex-M processors, even as a single task, command line only OS (since there is no ‘real’ MMU on Cortex-M, it’s probably an easier target).

A subset of RISC OS that could run BBC Basic and CLI tools.
RISC OS embedded? :)

Something that could run RISC OS Basic and CLI tools.

Similar to, but a bit more stand-alone than https://github.com/hoglet67/PiTubeDirect/releases ?

But why??? Do you have a customer in mind?

> The huge difference is the scale of what’s needed. A few very clever people
> was all it took to design the original ARM CPU, and the original RISCOS.
> The world has moved on, and to excel in today’s world things need to be
> a lot more complicated, and a few very clever people isn’t enough.

Meaning a company with a rather big engineering staff all experts at C/ArmV9 and nice big black numbers in their company budget profit column could do it.

Or since it’s open source, rely on a similar quantity with identical skill set from within the RISC OS tribe to write it as a distributed/demarcated collaborative effort under elected management. This scenario hasn’t appeared in the ~11 years since Arm went 64-bit, so what’s the likelihood of it appearing now? Since the first option is hypothetical at best, what is the likelihood of the Tribe doing it? Slim I’d wager, for reasons already stated.

And writing it is one thing, maintaining and developing it further another. It would take an established company structure (doesn’t exist) with a solid monetary surplus to do this, so unless this can be arranged a promising technology disappears. Not all at once of course, but without true development it will loose monumentum and become a historical footnote.

but without true development

Are you willing to cough up some coins to pay actual developers to do this work?

>> but without true development

> Are you willing to cough up some coins to pay actual developers to do this work?

Willing to if I had any, yes, absolutely. As it is I have no monthly surplus, and I’m not living extravagantly, far from it. Being Norwegian doesn’t automatically equal being stinking rich, you see.

After the referendum tanked the pound (which has pretty much been heading south since then)

My GBP/EUR chart tells a different story. Pound approx. the same value as in mid-1995. Higher than 2009. Rising compared to the “big dips” (which look a lot like speculation-induced artifacts) in 2019 and 2020.

As always with statistics: choose the right start date for the trend period you want to base your hypothesis on :-)

Looking at the chart, I am a bit puzzled about the pound high in mid-2015, but it was probably just a EUR low instead. But I don’t want to derail this thread further.

> But I don’t want to derail this thread further.

Why, thank you. :)