Acorn Arcade forums: What is your current RISC OS setup? (David Feugey)

What is your current RISC OS setup? (David Feugey)

This time, we pop across the channel to find out what David Feugey's desktop looks like...

I have two RISC OS sets.

The first one is in the office and is used for all of my business activities. A good 2560x1440 screen, with a perfect audio setup and a Pi4 @2.0 GHz running RISC OS 5.28. Under the desk, there is a box with a PC (Intel NUC), accessible from RISC OS via RDP. I plan to add another computer for development work. I also have an ARMbook, not visible here.

The second set is in the workshop. Photos are taken from left to right.

The first computer is behind the screen. It is a Pi4 @2.147 GHz connected to an AverMedia device, for video sessions. The PC on the left is a semi-functional laptop, without a screen, and now used to upload videos (and a few other tasks, for example emulation).

The second computer is actually a collection of computers. Iyonix PC , Pi1, Pi2, Pi3 and Pi4. A complete set of CPUs running under RISC OS 5.28. This is my test bench, which is now used to create the new RISC OS 5 software catalogue for the RISCOSFR website. I also use it to test my own software.

You can see two computers here. Both are useful for reading and testing floppy disks and old media. The first one is a StrongARM RiscPC, with a full SCSI chain. The second one is an Archimedes 410/1 with all the options (ARM3, 4MB RAM, SSD). Both have a MIDI interface (DMI 50). They are my music studio :)

The last one, on the right, is an Acorn A7000+. It doesn't work very well, and it's not very useful for things other than testing my own software. If only I could find a one slot backplane to install my ViewFinder inside it... I could drive a better screen and try to find new uses for this computer.

You can see lots of other RISC OS Computer setups

If you want to add yours, send us send a pic and an intro on your RISC OS related setup (email to markstephens At idrsolutions.com), and we will add it.

What is your current RISC OS setup? (David Feugey)

nytrex (10:16 9/1/2021)
dfeugey (22:15 11/1/2021)
markee174 (14:10 12/1/2021)

Alan Robertson

Message #125043, posted by nytrex at 10:16, 9/1/2021

Member
Posts: 117

Very cool. Lots of great RISC OS equipment at the heart of your house.

[ Log in to reply ]

David Feugey

Message #125045, posted by dfeugey at 22:15, 11/1/2021, in reply to message #125043

Member
Posts: 40

Thanks

Nota: the link of the first photo is broken...
Nota2: I'll have a studio soon for live capture.

[ Log in to reply ]

Mark Stephens

Message #125046, posted by markee174 at 14:10, 12/1/2021, in reply to message #125045

Does all the
work around here
Posts: 154

Thanks

Nota: the link of the first photo is broken...

Fixed.

[ Log in to reply ]

David Boddie

Message #125048, posted by davidb at 16:38, 12/1/2021

Member
Posts: 147

Nice to see the A7000+ get some love.

Are you running stock RISC OS 3.71 on it?

[ Log in to reply ]

David Feugey

Message #125049, posted by dfeugey at 17:55, 14/1/2021, in reply to message #125048

Member
Posts: 40

Yes. It was crashing a lot, and can't open a CD. After some investigations, I found it was because of the softload upgrade to RISC OS 4. So I simply remove it!

[ Log in to reply ]

David Boddie

Message #125050, posted by davidb at 15:14, 15/1/2021, in reply to message #125049

Member
Posts: 147

I never even considered an upgrade to RISC OS 4 because it didn't provide anything I was interested in, so mine still uses 3.71.

The power supply whined a lot the last time I used it, so I'm thinking of replacing it at some point. Do you have any tips on keeping it running smoothly?

[ Log in to reply ]

David Feugey

Message #125052, posted by dfeugey at 17:33, 18/1/2021, in reply to message #125050

Member
Posts: 40

The new filesystem coming with RISC OS 4 is great.

For the PSU, sorry, no tip... On the RISC PC, I have a classic PC PSU. Easier to replace

[ Log in to reply ]

David Boddie

Message #125053, posted by davidb at 21:33, 18/1/2021, in reply to message #125052

Member
Posts: 147

I was already using raFS, which was a neat hack and pretty robust.

[ Log in to reply ]

Andrew Rawnsley

Message #125054, posted by arawnsley at 15:12, 19/1/2021, in reply to message #125053

R-Comp chap
Posts: 600

As David said, the new FS is not only effective for long filenames, but much more efficient with hard drives (capacity/usage), removing most of the restrictions and bugs present in 3.7/3.71, in a way that remains compatible with modern machines (unlike RaFS).

On top of that, RO4 was significantly faster, especially on that class of hardware. We typically (if I recall) saw up to 30% speed gains, and routinely 20+%. That was significant on those machines.

I personally wouldn't want to use a RiscPC/A7000 without OS4, but I know some of the retro crowd prefer the "pure Acorn" aspects of 3.x warts and all!

[Edited by arawnsley at 15:13, 19/1/2021]

[ Log in to reply ]

David Boddie

Message #125055, posted by davidb at 10:44, 20/1/2021, in reply to message #125054

Member
Posts: 147

I don't remember anyone mentioning speed improvements like that at the time. That would have been interesting, but perhaps not applicable to some of the tasks I used the machine for. Maybe the magazines ran some benchmarks.

The promotional brochure for RISC OS doesn't compare RISC OS 4 to older versions, understandably. Some of the "features" listed were poorly chosen.

One interesting thing is that the brochure mentions various systems, including one I hadn't heard of before: the SVD Visiobus. Apparently, it provided electronic newspapers on public transport, presumably in France. Maybe David knows more about that...

[ Log in to reply ]

Rob Kendrick

Message #125061, posted by nunfetishist at 10:47, 28/1/2021, in reply to message #125055

Today's phish is trout a la creme.

Posts: 524

I don't remember anyone mentioning speed improvements like that at the time. That would have been interesting, but perhaps not applicable to some of the tasks I used the machine for. Maybe the magazines ran some benchmarks.

Most, if not all, of the performance gain was provided by "lazy task switching" (other operating systems call this demand paging). Basically when you switch between tasks, rather than remapping all of the task's memory, you only map in the pages of memory as they're needed. For apps with larger WIMP slots this is a massive win because most of the time you only need to swap in the pages involved in handling their Wimp_Poll loop. (So as little as a one page out of thousands!)

[ Log in to reply ]

Andrew Rawnsley

Message #125062, posted by arawnsley at 11:03, 28/1/2021, in reply to message #125061

R-Comp chap
Posts: 600

Actually Rob, I'm not 100% sure that's true (the "most if not all" bit - the rest is perfect). It is a long time ago, so my memory may be a little "off", but my recollection was that lazy task swapping had to be disabled on many systems since it was only supported by rev-T or later StrongArms which weren't super-common (I think we only had one in the office)

Whilst there *was* a performance gain from that, it was (mostly) fairly minimal.

I remember testing RO4 for performance in the office on several RiscPCs of different types, and I think we only had one rev T, yet performance boosts were very visible across the board.

But this is all more than 20 years ago, and I can't recall why else it would be so much better. But it definitely was, and it didn't matter whether you had a ARM610, 7500 or SA. They all seemed noticably snappier.

[Edited by arawnsley at 13:27, 28/1/2021]

[ Log in to reply ]

Rob Kendrick

Message #125063, posted by nunfetishist at 14:14, 28/1/2021, in reply to message #125062

Today's phish is trout a la creme.

Posts: 524

But this is all more than 20 years ago, and I can't recall why else it would be so much better. But it definitely was, and it didn't matter whether you had a ARM610, 7500 or SA. They all seemed noticably snappier

Demand paging will be a *huge* benefit to those slower CPUs. I'm not saying that there were not other performance improvements, but it would have been a major contributing factor. On StrongARM, ISTR somebody once saying there were also improvements in where cache flushes happen (not needed on earlier CPUs due to combined I/D cache) - in 3.7 they were rather scatter-gun, and improvements in using CallASWI instead of creating the appropriate SWI instruction (which requires a cache flush, and it means you end up using both the I and D caches for your instruction, meaning the data gets cached twice, which means less efficient cache usage.)

But I bet that lazy task swapping (assuming non-K-revision StrongARM) is of huge benefit in responsiveness and "context switches per second".

[ Log in to reply ]