Cloverleaf Campaign is Live

What is the scope of multi-cpu support? The pie in the sky version is RISC OS tasks seamlessly run on multiple cores, multiple threads can be used, each able to use all OS APIs – but that can’t happen without a complete re-write and a pre-emptive wimp. The more likely version is our current single core RISC OS, with a simple monitor which runs on the other cores, that sets up shared memory, runs some code which has either no access to the OS APIs, or at best very limited set of SWIs, then returns results back to the main core via the shared memory – that’s not much work, but not that much use either, and and several people have already achieved that in the past.

And I would have been a Phoebe 2100 owner :-)

If you’d been asked to pay anything?

Michael Gebracht – I agree with both halves of your post. I think it is probably best to think of the further RISC OS developments in terms of “stretch goals” given the low (for a hardware project) target. I know that Stefan was keen to try and make the initial target reachable because many backers will not back a project until it is seen to reach its target (chicken, meet egg!).

One of the reasons you’ll see my name in the project info is that I am keen to try and ensure that Cloverleaf works with the existing development. You’ll note that members of the existing scene are being supported by Cloverleaf – there are videos from Wifi Sheep, designs from RISC OS Bits, R-Comp/RCI is involved as are Archive and so on. I should clarify that I’m not a cloverleaf director or employee, but rather that I have tried to steer things and provide information so that Cloverleaf complements the work being done already by ROD and ROOL. My hope is that a successful Cloverleaf campaign will be a symbiotic success with/for the various RISC OS companies. Indeed, I think it will be a two-way street because Cloverleaf will need tech from the existing companies to deliver on its vision, and the existing companies will benefit in return. Well, that’s my hope, anyway.

@ Andrew Rawnsley

However, Kickstarter has earned big interest and money for other retro projects such as Spectrum Next and others, so why shouldn’t someone try a RISC OS one? Goodness knows our beloved platform could do with some more exposure and external interest :)

I am with you on most the things you’ve mentioned sir! But this one needs a bit of clarification:

• The constant comparison with the ZX Spectrum Next Kickstart:
  • The ZX Spectrum is one of the most sold micro computers in the history. The most sold Acorn ARM based computers were probably the Archimedes series (AFAIK and happy to be corrected if I am wrong), which were not significant to the history of Acorn so much that even the history of ARM (in many cases) still mention the BBC Micro instead…
  • The above creates a profound difference between the ZX and the retro ARM platforms, which may not have the same effect on kick starters as much as the ZX Spectrum Next had.
  • Also let’s not forget that the ZX Spectrum had nothing comparable to the ZX Spectrum Next, on every base (I am talking as a ZX Spectrum owner and coder here), the ZX Spectrum Next is definitely a whole new level for the ZX community that incorporates ALL the past and compatibility with the past and a whole new future with highest performance, WiFi, new high res modes, integration with ARM and obviously an extremely beautiful design. It also had an important piece of history the original ZX Spectrum+ designer behind the new case design.
  • While the Cloverleaf kickstart has nothing of the alike. But with that said I do not want to sound negative with the Cloverleaf project, which I believe if they’ll manage to achieve 25% of what they promise in terms of changes on the OS it will be a success anyway.

• on the “so why shouldn’t someone try a RISC OS one?”
  • I agree, someone should TRY IT regardless the above, why? To actually measure that love for RISC OS in 2020 and make sure we all understand that exact amount publicly, so we either get the whole pack of benefits from Cloverleaf project OR we’ll learn how to manage our decisions better for the future, so again best of luck to Cloverleaf :)

• On “Goodness knows our beloved platform could do with some more exposure and external interest :)”
  • As a retro platform I’d totally agree with you, as an IoT platform as well, but as a modern desktop… man I really hope RISC OS doesn’t get popular before it gets modernised for real… otherwise the backlash could label it as a crazy-idea for many years to come… but this is just my personal opinion so worth probably less than 0.5c

Just my 0.5c as always and I hope this comments make sense and in a constructive way, if not then apologies.

As a retro platform I’d totally agree with you, as an IoT platform as well, but as a modern desktop… man I really hope RISC OS doesn’t get popular before it gets modernised for real… otherwise the backlash could label it as a crazy-idea for many years to come… but this is just my personal opinion so worth probably less than 0.5c

I think perhaps 0.5c is wrong.

Try a dollar value.

@ Paolo

I agree with almost all of what you say.

However, I’d argue that for the many RISC OS users who left in the 1990s or early 2000s (I spoke to one returning gentlemant today, for example), then modern hardware/software does represent that kind of generational jump (especially if legacy compatibility elements are included. That being said, the case design of Spectrum Next is quite special, and I think we’d need a lot of investment to be able to offer something similar for a next gen RISC OS box. However, I personally hold no particular affection for the single-box computers (BBC, A3000) despite growing up with them – I always saw the separate component look of the Archimedes / A5000 / RiscPC as aspirational. That said, I’m now going misty-eyed thinking of my A3000, and am typing this on a keyboard with red function keys, so…. touche!

I’m (very) skeptical of the timeframes too. However, Stefan and his programmers have delivered on ChatCube and it’s really rather impressive. I know that it’s a huge jump from an instant messenger to some of those Kickstarter goals, but it’s still some sort of a proven track record. I wish him well and hope he surprises us all and is successful.

And I would have been a Phoebe 2100 owner :-)

If you’d been asked to pay anything?

You mean like the Clan advance order and matching discount offer or CTA’s £100 deposit offer.

Anyway back on point..

I believe if they’ll manage to achieve 25% of what they promise in terms of changes on the OS it will be a success anyway.

Just as long as it is the Sound/Bluetooth/Wifi bit and not multi colour dirtectories as part as part of the update GUI look :-)

The more likely version is our current single core RISC OS, with a simple monitor which runs on the other cores, that sets up shared memory, runs some code which has either no access to the OS APIs, or at best very limited set of SWIs, then returns results back to the main core via the shared memory – that’s not much work, but not that much use either, and and several people have already achieved that in the past.

I 100% agree. And that’s not much work… since Jeffrey already done it. It would be cool to have a ROM with this feature available and a few C and ASM examples. Basic examples for me :)

since Jeffrey already done it. It would be cool to have a ROM with this feature available and a few C and ASM examples.

Well there is still work to do there, for instance the latest release still doesn’t support VFP and Neon on the other cores.

Examples are actually available and I already started to work on it for my AI/ML stuff, but then I stopped testing on it because actually right now a Neon based math library would make more sense to reduce Genann latency more, so I am looking on that direction at this time.

https://gitlab.riscosopen.org/RiscOS/Sources/Programmer/SMP/-/tree/master/Test/

I think Jeffrey’s work on the multi-core is impressive and I really like it so far, so really hope that the Cloverleaf project will push it further :)

I’m now going misty-eyed thinking of my A3000, and am typing this on a keyboard with red function keys, so…. touche!

:))))))

Ok here is my proposal… Raspberry Pi 400, overclocked and with green function keys!!!! …and then runs away and hide! :D

with green function keys!!!! …and then runs away and hide! :D

Pop along to your local general store and buy some of the nail polish suitable for Emo/Goth use and you can have your own shade of green (or whatever)
Then explain that to your friends :)

Well not exactly new given that VirtualAcorn made a business on the back of Red Squirrel if my memory serves me right

Oh, now there’s a Trumpian Truth.

Yes, I believe that VirtualX did develop from what was originally RedSquirrel, however an awful lot of work was put into the VirtualX systems in order to greatly improve the speed and capability of emulation, such as JIT.
I don’t have the emulator, so I’ll let somebody else point out the enhancements.

so should they be used in the kickstarter timeline?

It depends. Are the people already involved going to be taken on in a more “paid work” sense by Cloverleaf, or is this something else?

When I mentioned March 2021 for multi-core, it was because of a mixture of surprise and scepticism.

Ditto. I can only imagine that RISC OS, as it stands, will use other cores in a Tube like fashion. There are numerous API issues involved in spreading tasks among cores (for instance, no concept of EWOULDBLOCK; and consider how one would cope with UserMessage_Recorded).

Put simply, if it was easy to spread RISC OS across multiple cores, Jeffrey probably would have already done it by now. ;-)

The reward will rarely be “worth” the amount directly, simply because the money is meant to “kickstart” the idea/project.

Actually, it seems to be quite common that the object being offered is offered for less than the RRP, or with extras. Basically as a thank you for trusting them, being willing to take the risk, supporting them, and letting them demonstrate that they aren’t offering a vapourware Sinclair Vega+.
The kickstarter helps to raise the cash to “make it happen”. The profit comes from the mass production and sale of the object.

I also note that this will not even happen until the threshold is reached on a All or Nothing basis.

Well, they’ve already raised ~15% from 39 people in one day. 58 days to go to find a further ~€42K. It’s a large amount, but there’s time…

and I am curious about what “updated GUI” is about.

Rounded icons.

<ducks and runs away>

instead of going out of its way to obfuscate.

Yeah, there’s a hell of a lot of blahblah there, it’s quite a lot to take in. There is also a lot that is rather subjective – “robust OS”, some waffle about carbon footprints, a flawed calculation in determining that the processor is 225 times faster ¹, something dubious about forced upgrades ², something about incompatibilities (hmm, unaligned loads, SWP, zeropain, etc etc), and so on.

Duke Nukem is available for RISC OS? That passed me by…

It’s also interesting to read about the desire to give everybody an alternative to the three monopolists that only want to line their shareholder’s pockets. Well, uh, the problem is that there were alternatives that basically got swept by the wayside. Anybody remember Symbian? Or Blackberry? Or, hell, a phone running Windows? Even the mighty Microsoft managed to bugger that up!

fear that the timescales are unrealistic and therefore the goal may not be achievable.

I’m looking forward to next August – GPU video decoding. When RISC OS is able to let people watch kitten videos on YouTube (in HD), that’s what will make the difference.

If I can advance a suggestion for 2022 (seeing as 2021 is spoken for), it would be nice if there is support built into the system for IPP Everywhere or AirPrint. Pretty much every WiFi compatible printer supports some version of AirPrint. The more recent models also support IPPE. Having this would be a good thing, as it would allow fairly effortless printing to modern printers without all the rubbish of sorting out drivers and the like. My iPad Mini, for example, has a print system aimed at morons (pretty much the only option is “one copy or more?”), but it worked with my laser and inkjet with no setup whatsoever. It found them, and spat out something. Okay, for some reason it filled my test (google’s intro page) to fill the entire page, but other than being a bit dumb, it worked.
[I still have nightmares over trying to set up WordPerfect 5.1 on the office computer to talk to some oddball laser that only seemed to understand Epson LQ (yeah, DOT MATRIX!) commands; it did not support HP LJ2]

How will this pay for all the development and programmers?

What’s the going rate for an experienced programmer that speaks assembler and not some trendy modern paradigm?

I should clarify that I’m not a cloverleaf director or employee,

The page sort of suggests otherwise.

Quote: The main figures behind this project are Stefan Fröhling (former Evolution Computer GmbH, Germany) and Andrew Rawnsley (R-Comp Interactive, United Kingdom).

And a little further down: Stefan found a competent partner in Andrew […]

I am keen to try and ensure that Cloverleaf works with the existing development.

Well of course. We’ve seen what happens when two incarnations go their own way. Much better to have everything in sync.

The constant comparison with the ZX Kickstart

Yeah. I’ll take your Kickstarter ZX Next and I’ll raise you an Indiegogo Vega+.

Are you running screaming for the hills yet? ;-)

It also had an important piece of history the original ZX Spectrum+ designer behind the new case design.

<shrug>
It’s what’s inside that matters… ³

otherwise the backlash could label it as a crazy-idea for many years to come…

Well, for that to happen it would need to first become popular. The year of RISC OS on the desktop? :-p

then modern hardware/software does represent that kind of generational jump

Oh, yes. So very much this.

To the point where, so killer app excepted, I really don’t understand those who stay stuck to their aging dying RiscPCs.

I mean, a computer that was being sold before Buffy that is a big clunky desktop machine, with either an ass-kicking fast processor bolted to a tragically slow memory bus and a maximum of 2MiB VRAM (or in some instances, an ass-kicking processor and associated memory, but absolutely slaughtered I/O),,,

…versus a little box with nippy gigahertz speeds processor, easy use of modern USB storage, a gigabyte or more of memory, and no trouble talking to a monitor with an insane number of pixels. Oh, and it can run for a fair old while on one of those little battery packs that you can charge phones with.

There’s no comparison really. It’s more than a generational jump, it’s a run-skip-and-jump into the real world.

That said, I’m now going misty-eyed thinking of my A3000,

Got one in a cupboard somewhere? Try using it.

Amy was my first RISC OS machine, and my second Acorn (after a Beeb). It was a massive leap forward and I did a lot with that little A3000, including fitting Econet, patching into IIC to run a teletext receiver, and even stuffing in a dinky harddisc just about large enough to hold two songs (or one Nightwish song).
But going back to Amy after using something newer, as mentioned above, I was blown away by how SLOW it felt. Like, how much of my life did I waste waiting for this machine to do something? ⁴

typing this on a keyboard with red function keys

Lying on bed typing this on a Bluetooth keyboard with a simple Android tablet running Firefox. Well, it’s ARM based so I’m halfway there. ;-)

I wish him well and hope he surprises us all and is successful.

Me too, on all three counts. After all, his success will feed back to our OS so it’ll be a success for all of us too.

¹ That figure is derived by dividing the clock speed now with the clock speed then. However it is flawed because caching, speculative execution, and many years of architectural improvements mean that the current ARM processors can get more done in less time. Or, to put in another way, the perceptual speed increments of the various Pi models do not reflect their change in clock rate. My Pi2? Feels a fair bit nippier than the extra 200MHz would imply, but that’s because the Cortex-Awhatever is a much better processor than the ARM11. With that in mind, I would imagine that a proper comparison of the actual difference between the RK3399 and an ARM2 would make your head explode. I mean, jeez, when I went on holiday to Spain many years ago, I left my A5000 at home and took the old A3000. W..h..i..c..h.. f..e..l..t.. l..i..k..e.. t..r..e(oh FFS do something already!)..a..c..l..e.. i..n.. c..o..m..p..a..r..i..s..o..n..

² I’m still running XP on my PC box. While stuff has more or less stopped supporting XP, it had a really good run, and the software that I have now still works. Okay, I can’t let it loose on the Internet, but then, it’s XP, hasn’t that always been the case? ;-)

³ I have the same opinion with regards girls. A lot of people think that’s weird. I used to think it weird that guys would concentrate on the Gainaxing and not the attached girl…

⁴ Nothing compared to the amount of life wasted writing rubbish on forums!

Then explain that to your friends :)

Looooooool :D

Looking at the site now with a better browser I see the target is 50,000 euros. How will this pay for all the development and programmers?

By employing developers outside the Western world. There are rather a lot of them, often talented and up for a challenge. A largely untouched resource in RISC OS world.

IIRC, the developers who have done ChatCube had no prior knowledge of RISC OS. So it IS possible. However, it gets harder the nearer you get to the OS.

now there’s a Trumpian Truth.

The 21st century and most modern and hurtful of insults to be compared to that narcissistic bully is even worse than being sent to coventry or being called a sxxb in the 70’s.

Good job I took it in the nature it was intended then or was it :-)

2022 (seeing as 2021 is spoken for), it would be nice if there is support built into the system for IPP Everywhere or AirPrint. Pretty much every WiFi compatible printer supports some version of AirPrint. The more recent models also support IPPE.

Well you offering to improve SammyPrint then as part of CloverLeaf

Rounded icons.

damn, I just finished the “modern” flat/square desktop… :( lol

Quick unrelated question, does anybody would like to review a “modern” theme for RISC OS?

@Rick

Yes, I believe that VirtualX did develop from what was originally RedSquirrel, however an awful lot of work was put into the VirtualX systems in order to greatly improve the speed and capability of emulation, such as JIT.

That work was done prior to Mr Timbrel’s involvement, I beta tested the JIT on the final development versions of Red Squirrel. That was about the same time I wrote a network stack pass through to Windows for RedSquirrel which is suspiciously like what ended up in Virtual Acorn.

Apart from a rebrand and a price tag being added, it’s been milked for 20 odd years with bugger all ongoing development.

Quick unrelated question, does anybody would like to review a “modern” theme for RISC OS?

Yep.

Examples are actually available and I already started to work on it for my AI/ML stuff

Is a special ROM needed, or just some module?

Rounded icons.

Better hide too – JF’s reading :)

I’m looking forward to next August – GPU video decoding. When RISC OS is able to let people watch kitten videos on YouTube (in HD), that’s what will make the difference.

Kitten videos are where it’s at

If I can advance a suggestion for 2022 (seeing as 2021 is spoken for), it would be nice if there is support built into the system for IPP Everywhere

That’s probably easier to implement than most of the 2021 list.

or AirPrint.

First get reliable WiFi stack.

What’s the going rate for an experienced programmer that speaks assembler and not some trendy modern paradigm?

The “speaks assembler” clinches it, payment is in pizza and chocolate hob-nobs

Quick unrelated question, does anybody would like to review a “modern” theme for RISC OS?

This come with a dark & high contrast setup?

That was about the same time I wrote a network stack pass through to Windows for RedSquirrel which is suspiciously like what ended up in Virtual Acorn.

In an odd 5 minutes you could repeat that for RPCEmu?
(there are still some elements lacking in the Windows networking)

Yep.

Ok will put some pics on my blog and share on ROOL Forum then, thanks for the help in advance!

Is a special ROM needed, or just some module?

The SMP library is distributed as a module OR as source code to be recompiled (you’ll find latest source in the same link as for the examples.

HOWEVER (lol there is always a but) the pre-built module should be loaded/used only an RPi (i think RPi 3 probably is best) because it seems to be built for the Pi (for instance doesn’t work on my i.mx6).

Hope this helps David

Paolo – bit confused by your comments. An SMP module? Doesn’t work on imx6? I’m confused because AFAIK the only platform to have a shipping multi-core test build for developers is imx6 (I could be wrong here). And in that case it is a specially compiled ROM, not a separate support module.