RPi5 out

It [the Pi 5] is a long way from a low power processor running a light weight operating system.

Their products range from the Pico, the Zero 2W, the older 3A+, 3B+ and 4B Pi’s are still available, and now the most powerful Pi 5. A broader range of products isn’t a bad thing, I think sales would be pretty slow if they were still only selling warmed over Pi Bs (particularly as my original 256MB and 512MB ones still work).

I think sales would be pretty slow if they were still only selling warmed over Pi Bs

Yes, their problem (in their main market) is not having faster products than the Pi 5 – and such processors would probably use even more power.

The direction of travel is using more processing power to run less efficient software.

Computing might not have gone that way, been lots of worry over low power light bulbs but not low power general computing.

Part of it is that things went in a different direction than their original plans, that of low cost school and hobbyist devices, and a whole new market appeared and now needs feeding.

Reminds me of at least one encounter where someone was adamant that the Pi required additional extra features so that their brilliant business plan could take off. Needless to say the individual wouldn’t listen, utterly committed to not actually designing and developing their idea and wanting to exploit the Pi on the cheap. The Pi still does not have what they demanded.

things went in a different direction than their original plans

Good plans are often derailed by reality.
Well done them for spotting the trend and adapting/going with it.

The Pi still does not have what they demanded.

What was it they wanted? Was it something semi-reasonable like SATA, or something ridiculous like an interface for a third party GPU (because gotta mine that free money)?

They wanted HDMI in, their business idea involved a plan to build some sort of video processing device on the cheap.

Looking at the spec of the Pi 5, it would probably make a decent server for a small business, especially given the m.2 SSD compatibility.

8GB Pi 5 running Samba as an AD domain controller, Exim, cyrus-imapd, few other daemons, it looks like the perfect solution to running a back office server for a dozen users or so.

As for running RISC OS, there’s apparently no plans to discontinue the Pi Zero and Zero-W. I hope not anyway as I’ve got a few projects (not RISC OS based sadly) that use the Zero-W.

HDMI in

Yeah, okay, I can see why that was never taken seriously.

Re : discontinue

Pick a board and look at the bottom of the page for the “Obsolescence Statement” See CM4 for instance https://www.raspberrypi.com/products/compute-module-4/?variant=raspberry-pi-cm4001000 “Compute Module 4 will remain in production until at least January 2031”

CM4 is a nice opportunity for something custom at least.

No point to fit the SR71/Blackbird Pratt & Whitey rocket engines to the flying contraption built by the Wright brothers. Foisting an Operating System onto a CPU architecture which is 8 generations beyond the architecture RISC OS 3.6 was (to maximize speed but lock the code to a special CPU instruction set – not future-proof) assembly-coded for. Emulation must be written anyway, since otherwise nothing current would run on 64-bit RISC OS “as is”.

The key here is the 64-bit transition bringing RISC OS into the 21st Century (somewhat on the late side). I’d like it to survive since it’s the most productive (for desktop publishing/illustration) system I’ve ever used. Modern CPU/GPU combos are so much beyond the architecture current RISC OS was written for that even emulation (of old software) would be very fast indeed.

Now they released more information, especially on that auxiliary chip they developed for this board. It says it has a dual core M3, used also for startup but I guess that won’t help us either ? I’m no hardware person at all, just wonder how this one interacts with the big core…or just some i/o without caring of Thumb/32/64 Bit.

The part of the datasheet that caught my eye was “GPIO pins are 5V-tolerant”.

just wonder how this one interacts with the big core

There appears to be shared SDRAM (for running code?) so maybe something can be dropped in and the device asked to run it?
Thing is, what do they do normally? I mean, if their boot ROM stuff is for handling core functions of the chip, probably best left alone…

It’s at this stage 32-bit mostly assembly-coded RISC OS goes from being retro to outright obsolete.

Writing operating systems in assembler back in the day CPUs were relatively slow compared to current CPUs made sort of sense with the 4MIPS single-core Arm2 but is pointless today where an OS written in more portable C code can be relatively easily be ported from 32- to 64-bit CPUs – not “paint oneself into a corner” by using architecture-locked assembler. With multicore and multithread support it’d hardly be slower.

Anyone got one yet? Pimoroni said the cut off for the first batch of 8GB was at 23:59:59 on 28-Sep-2023, and of course I ordered a couple of minutes past midnight.

Since ‘Sveinung’ is keen – I wonder if Linux RO could be modified to work?

Linux has got its own raft of problems/bugs, the one good thing about it is the availability of 64-bit kernels/programs.

Risc OS were written as a separate platform, a status I think should remain since its user interface is as productive as it is. But staying 32-bit will not ensure its survival i the long run, only a 64-bit system with rock steady 26/32-bit emulation of legacy software can ensure that.

It’s a major undertaking since apart from porting the OS, a brand new 64-bit capable software development kit must be developed simultaneously. It could possibly have the blessing of using new graphic library standards enabling new functionality in new or converted RO design software.

Then it’d remain a separate computer platform, updated to modern standards. Maybe use .svg for icons so the icon bar’s height could be halved?

> Anyone got one yet? Pimoroni said the cut off for the first batch of 8GB was at 23:59:59 on 28-Oct-2023, and of course I ordered a couple of minutes
> past midnight.

28-Sep-2023 (not Oct).

In the UK, Rapid electronics currently have both the 4GB and 8GB RPi5 in stock for next day delivery. Look under new products.

I also ordered an 8GB RPi4 from Pimoroni (09:00 on 28Sep2023) but it has yet to ship.

Ralph

(typing this on his new RPi5 from Rapid. Very impressive in speed vs. the RPi4 with Raspian :-).

Come to think of it, why on a RISC OS forum discuss hardware on which RISC OS cannot run (without a non-existent sand-boxed abstraction layer)? It’s here at this stage reality transforms into pure fiction.

Has anybody calculated the number of man-hours and what expertise is needed to write this abstraction layer versus rewrite the RISC OS microkernel, its relocatable modules and everything else in C? What’s the logistics here? Given that the abstraction layer is a blatant short term stopgap solution for a very good operating system in dire need of a revival.

It might run RPCEmu quite well ;-)

Come to think of it, why on a RISC OS forum discuss hardware on which RISC OS cannot run

RISC OS Linux runs fine on the RPi 5. Just had to add kernel=kernel8.img to config.txt to use the 4k page size kernel. The default 16K one can’t run 32bit programs.

RISC OS Linux runs fine on the RPi 5. Just had to add kernel=kernel8.img to config.txt to use the 4k page size kernel. The default 16K one can’t run 32bit programs.

Good to know, Chris. I thought the large page size kernel might be a bit off for legacy stuff.

If RISC OS need to be run from Linux, RISC OS is dead as a Dodo as a separate computing platform. 64-bit RISC OS written in C would still out-perform almost everything else except maybe for floating-point stuff*. And be able to run legacy software under emulation at close to StrongARM speed, probably**. Thus it’d stay a separate (!) computer platform.

You may bitch and bark all you want against this view, but that’s the direction this hen is currently kicking – from retro to irrelevant. And hacks can only prolong RISC OS’ life for so long. Wwith a dwindling user- and developer base.

• Unless RISC OS is written to support modern Arm GPUs. “Stonking fast”, yep.

• Since RISC OS is to Linux like the hare to the snail on a tarried stick.

64-bit RISC OS written in C

There’s an excellent interview with Eben Upton in the latest Archive and much of the talk is about the 32 bit RISC OS problem. And of course it gets one thinking again.

Maybe if we just change the name, from “RISC OS” to “26/32 Bit ARM OS”, you will see the problem and that there is no solution short of “Another OS Altogether”.

If RISC OS need to be run from Linux, RISC OS is dead as a Dodo as a separate computing platform

What about running RISC OS in a web browser.

An 8GB RPi5, ordered yesterday from “Rapid electronics,”:https://www.rapidonline.com/raspberry-pi-5-4gb-8gb-589267 has arrived. (Too late, the word has obviously got around as they now appear to be out of stock.)

The RPi5 is impressive and noticeably livelier than the RPi4 and RPi400.

RISC OS for Linux on the 2400MHz RPi5 is faster than RISC OS on the 2400MHz RPi400.

It might run RPCEmu quite well ;-)

Peregrine falcon v Dodo. ;-)