September news round up | Drobe.co.uk archives

September news round up

Published: 18th Sep 2006, 01:07:54 | Permalink | Printable

RISC OS Open, XScale, RISC OS school, and show news [Updated]

It has certainly been quiet at Drobe HQ recently, as you've no doubt noticed. In between the holidays and increasing pressures from day-jobs, there have been a number of projects keeping us amused. While we wait for normal service to resume, here's some RISC OS news you may have missed.

Steve Revill to meet his public
7th Software and RISC OS Open staffer Steve Revill will be appearing before the London usergroup this evening. While he will primarily be showing off his MoreDesk software, Steve reportedly suggested to one contact that the launch of RISC OS Open would be "done and dusted" by the time of his appearence in the capital.

ROUGOL organisers told punters over the weekend: "Steve worked at Acorn, Element 14 and Pace from 1997 to 2002, and has also worked for Pace and Castle and Tematic since then. He will be demonstrating MoreDesk, as well as some of his other software, and we hope that he will also be able to talk a little about RISC OS Open Ltd after the main presentation."

Steve however played down the RISC OS Open aspect, adding that tonight's talk was planned before the new company was formed.

He told us: "I'm afraid there aren't any announcements on the RISC OS Open front to make just yet because we've been stuck in various discussions with third parties, sorting out various bits of planning and getting legal agreements in place.

"There are some interesting things happening next week so we hope to have something to announce very soon now."

Incidentally, Drobe's Chris Williams is pencilled in to appear next month at the usergroup. The organisers have billed the evening's entertainment thusly: "Drobe has provoked much controversy and criticism from different RISC OS companies, most recently RISCOS Ltd. The evening should also give you a chance to hear some of Chris's own personal opinions about matters in the RISC OS world."

Intel XScale big in Japan
A classy source clearly on the ball recently tipped us off via his Blackberry that Intel's new monster 1.2GHz XScale core is codenamed Chevelon, and has been demonstrated as part of a RAID card. Earlier this year in Tokyo, the chip giant also touted the ARM-compatible processor on an ATX-style reference motherboard with a PCI Express socket, a PCI-X socket, two serial ports, two gigabit ethernet ports, CompactFlash expansion slot, and a JTAG programmer.

Knightsfield school

RISC OS school for blind seeks more staff
IT award winning school Knightsfield is looking to hire a maths teacher. Deputy headteacher Sharon Pointeer said they hope to be taking on teaching assistants too. We earlier helped the school fill another role at the RISC OS powered school for blind children.

She said: "We are advertising again - this time for a maths teacher. There just has to be people out there who are RISC OS enthusiasts and maths teachers surely?

"We have a really lovely school, our pupils are great, and well behaved, and the school has such an excellent reputation that we are over subscribed."

New kid on the block
A new bi-monthly magazine backed by Paul Vigay and set to fill in the gap left by comatose Qercus is taking on subscriptions. RISC OS Now editor and trainee secondary school maths teacher Louie Smith said she hopes to get the presses rolling by September 29, and the completed magazines ready to be posted by October 9.

She said: "At the moment, I do have some great content varying from very basic stuff to more high-brow reading. This magazine will not work unless people are prepared to tell me what they want and support it. I don't want the violins, but I will be working full time as a teacher, reading for my PGCE degree, and producing this magazine, so I need positive and constructive criticism."

For more information, click here or read this, but ignore the bit about the cheque guarantee numbers. Meanwhile, ARM Club committee member Dave Ruck has called for John Cartmell to be banned from future RISC OS shows as he has been mysteriously unable to publish a single Qercus issue since around this time last year.

Show news

South west show wastes no time
Next year's RISC OS show in the south west of England is set to take place on Saturday, February 24. Forward thinking organiser John Stonier said the event is booked for the usual venue of the Webbington Hotel, north Somerset.

He added: "Hopefully, most of the main RISC OS companies will be there, mixed with hobbyists, groups, and the usual games and charity stands. The theatre presentations will offer an opportunity to expand your knowledge and keep abreast of RISC OS developments. Please support the show once again and show us that the RISCOS scene is very much alive."

Nostalgia trip a day keeps the doctor away
As pointed out by ANS, a number of scanned copies of Acorn's internal company newsletter have surfaced online. These historic issues of Acorn News will jerk tears from even the more cynical of Acorn users, as the corporate propaganda celebrates Acorn's golden era. Intelligent Interfaces have also published user guides of their legacy podule expansion cards online as PDFs. Manuals are now available for their IEEE488 GPIB card, 16 bit parallel IO device, STEbus interface, dual high speed Serial card, single width IO interface, and 12 bit Irlam Instruments sampler.

And finally
A free update for Easiwriter and Techwriter users from version 8.60 to 8.63 is available for download. Chris Wraight penned a funky article about RISC OS for OSNews, and says he will release the gorgeous desktop theme he designed and used in the article's screenshots. Open source speech synthesiser Speak is now up to version 3.03, and the experimental ELF GCC package has been updated. Paul Vigay has also created a new utility to convert Fireworkz files into web pages.

STD have axed their 100MBit ethernet card four years after it first went on sale. Back in 2002, one astonished design engineer working for STD joked: "The cards are flying off the shelves, we've sold more units than we thought there were RISC OS users left." Now STD say "difficulty with supply of key components means that the manufacture of further stocks of NET100 interfaces is unlikely". The device will be superseded by the Unipod, which has an optional 100MBit ethernet interface.

Update at 15:08 21/09/2006
Steve Revill was keeping tight-lipped about RISC OS Open at ROUGOL, other than to say the company is sorting out legal issues with unspecified people. He said that RISC OS Open would be "good for the platform", and while their work is related to open source software, punters at the event got the impression that Steve was not refering to the actual operating system itself. Steve added that, while he was at Pace Micro, the programmers did the 32-bitting of RISC OS without the approval of their management.

Links

What have you got to lose? Give us your news

Previous: ROS fan loses Acorn domain in dispute
Next: GCC 3.4.6 in bug fix release

Discussion

Viewing threaded comments | View comments unthreaded, listed by date | Skip to the end

Hmmm, so there's an X-scale ATX motherboard out there. This would make sense in todays world that is lowering the power consumption of products. We already have X-scale PDA's with mobile windows on it. So this would suggest that a fully fledged "normal" windows could be achieved and maybe already out there under test.

How sad would it be to see Windows Vista running on an X-scale machine but no RISC OS alternative. Even sadder: such a machine would be marketed as a new generation of low power PC's which steals some of RISC OS's limited thunder.

mripley on 18/9/06 8:26AM
[ Reply | Permalink | Report ]

mripley:

ATX motherboards exist for a variety of ARM processors. The reason for this is usually to provide a development platform for testing software on real processors. Have a look on simtec's website.

Have you seen Vista? It stuggles to run on a P4, the X-Scale has no chance!!

X-Scale processors can be found in a variety of RAID controllers. This isn't anything new either. The processor used in the Iyonix is more commonly found on RAID controllers.

knutson on 18/9/06 10:17AM
[ Reply | Permalink | Report ]

This won't happen. There have been ARM-powered (ATX-) mainboards available for just about every ARM ever released, simply because embedded hardware developers like to use them for prototyping. They are usually sold as developer kits. This should make it a lot easier for Castle to make an Iyonix2, though. :-)

JGZimmerle on 18/9/06 10:19AM
[ Reply | Permalink | Report ]

@knutson: I guess I should have pressed the send-button before getting breakfast. :-)

Anyway, you are right about the RAID controllers. I have a more powerful XScale on my PC's RAID controller, than any native RISC OS machine as its main processor. :-(

JGZimmerle on 18/9/06 10:23AM
[ Reply | Permalink | Report ]

It interesting to see that Chris will be defending his comments/views on ROL at the next London usergroup.

Personallly I think he has every right to share his views whatever they are. PM says that open sourcing won't work because of the costs associated with it, but lets looks at ROL.

PM salary - Â£30k programmer salary - Â£30k overheads - Â£5k

I'm being conservative, but thats Â£65k a year that we have to fund for ROL just to break even. I think Open sourcing has a chance!

So C'mon Chris share your views. Don't be afraid.

nx

nx on 18/9/06 1:00PM
[ Reply | Permalink | Report ]

The usual games and charity stands? There was no games stand at SW2006.

CheatWarrior on 18/9/06 1:14PM
[ Reply | Permalink | Report ]

Steve Revill to meet his public:

If anyone is thinking of trying out a virtual desktop program may I heartily recommend Steve Revill's MoreDesk. It is well worth the fifteen pounds.

I have been using Dave Ruck's Workspace for quite a while but was intrigued by reports on MoreDesk which I bought last week. Workspace is an excellent program and I have used it daily for over a year. It is free and has a powerful set of commands for setting up your desktops, but it has some drawbacks, chiefly the lack of an overview window where you can see your all open windows and, for me, the inability to change the set number of virtual desks.

MoreDesk, has a very good overview window which uses icons to represent your open applications. It also neatly solves the problem of having some applications show on all virtual desktops. You can have up to 49 desktops (I believe), but I am currently using three, and will probably add more when I get comfortable with three and need more space.

A good reason for buying MoreDesk is the quick response feedback. Last Thursday night I sent an email with details of two problems. Just before midnight Steve sent back a bypass fix for one and managed to include fixes for both problems in the new version 1.10 available two days later.

JohnR on 18/9/06 1:31PM
[ Reply | Permalink | Report ]

In Reply to JohnR

"I heartly recommend MoreDesk"

After John's excellent demonstration of MoreDesk at last weekend's MUG meeting I sure that Steve will get a few more sales of his application.

As to the bit about the 1.2Ghz Xscale then i do hope that it sees the light of day on RISCOS sooner rather than later. What would be even nicer is some way of utilising the 2nd core to run Linux and have this as a slave to run applications that we can't run on RISCOS but make them appear as running in a normal RISCOS window, if that makes sense. Likely candidates would be any Video streamer/web browser, flash etc i.e the usual suspects that would make RISCOS even better.

bluenose on 18/9/06 5:12PM
[ Reply | Permalink | Report ]

bluenose: Hmm, I'm not really an expert in this sort of thing, but your suggesting sounds annoying implausible even to me. You'd need too modify both OSes so they could share the memory, hardware (such as HD controllers) without interupting each other, so that's a customised version of Linux specific to that hardware. A customised X server to output graphics info accross a 'new magic bridge' between the two OSes, the same bridge would have to pass keyboard/mouse events back to the linux half.

All of this may be possible with huge amounts of money. But still it would be cheaper for RISC OS users to buy a standalone x86 linux box, that runs so much faster than a 1.2GHz ARM box, for just the same applications.

flibble on 18/9/06 6:00PM
[ Reply | Permalink | Report ]

In reply to Fibble

Thanks for the reply, we can only dream and as a none expert myself thanks for not being to hard on me.

I would like the apps so perhaps it's back to the arguement about funding/co-operation again for ported/native apps.

bluenose on 18/9/06 6:37PM
[ Reply | Permalink | Report ]

bluenose: "funding/co-operation"

I am delighted to report that Steve sold out of all the MoreDesk CDs he brought to ROUGOL, and also (I think) one or two extra sales beyond that. A very well received presentation!

Co-operation was also much in evidence after the meeting.

bluenose, you say there is an argument about funding/co-operation. So, what are you arguing in favour of?

dgs

dgs on 19/9/06 2:26AM
[ Reply | Permalink | Report ]

Well, first of all, thanks to all who attended the meeting - I had a great night with lots of good feedback and ideas. It was well worth the trip down to London.

Secondly, I did have to smile when I saw the bit in the article above which quoted me as saying the RISC OS Open Ltd launch would all be "done and dusted" by the time of the talk. To put my quote back into context, I said it to the talk organiser (dgs) back in July, when I had high hopes of speedy progress (little did I know). For the record, here's what I actually said:

"I suspect it'll all be done and dusted by then as far as launching the company is concerned. I probably won't mind talking a bit about it, but after talking about MoreDesk because otherwise it'd hijack the talk."

Sadly, things didn't move quite as quickly as I'd hoped - and they never do when you've got legal work involved - but I am still hopeful that there will be something positive to announce in the very near future.

7thsoftware on 19/9/06 2:46AM
[ Reply | Permalink | Report ]

In reply to DGS

Well ideally I'd like to see some form of collective scheme to drive applications forward like Select did in the early days of the scheme. This would be used to fund the best way to get the usual apps requested on to RISCOS be it a native version or port.

I'd still like to think that the operating system fork could be converged and that future development concentrated on RISCOS4/5 as a minimum. Eventually the RISCPC format will have to be left behind in favour of A9's/Iyonix and VirtualAcorn developments.

How all this is to be achieved of course is the $64 question. Still we can dream but as a die hard RISCOS fan I know that this needs to be done soon before the critical mass shrinks any further.

bluenose on 19/9/06 8:04PM
[ Reply | Permalink | Report ]

Whilst the reality is that the AT-X board with this new chip exists soley for the benefit of PDA manufactuers, it is nice to know that such a board is technically possible. That putting one of these new Intel chips in a desktop machine is not just the pipedream. Of course getting RISC OS to work on one of the cores is a greater technical challenge, and getting the old Hydra API to work so that both cores can be used is a pipedream.

JWCR on 20/9/06 3:30PM
[ Reply | Permalink | Report ]

Carefully reading Chris' article one *might* note that the ATX board spoken of was mentioned as having "a PCI Express socket, a PCI-X socket, two serial ports, two gigabit ethernet ports, CompactFlash expansion slot, and a JTAG programmer", that's *not* the fare of a typical PDA.

As to getting it to work much of the groundwork (HAL, RO5/32bit) already exists. The processor is pretty much like the 80321 (albeit with two cores, extra cache etc.,). If it is simply used as a *single* processor CPU then chances are the *existing* RISC OS 5 will work.

Yes we *won't* be able to use the 2nd CPU on the die because of RISC OS limitations - but there *is* something to be said for using a processor with a large L2 cache, twice the clock speed, support for faster interfaces (like PCI-X/Express) that should give somewhere around x2-x3 times the performance of the Iyonix 80321 even when *only one of the processors is used*.

AMS on 20/9/06 6:24PM
[ Reply | Permalink | Report ]

AMS: we wont be able to use the second core for running RISC OS applications, but there is bound to be significant processing tasks that can be off loaded on to it, leaving more of the the first core's time free. We do have the histrory the Beeb 2nd processor systems to draw on for inspiration.

druck on 21/9/06 9:05AM
[ Reply | Permalink | Report ]

druck "We do have the histrory the Beeb 2nd processor systems to draw on for inspiration."

But processing on a second processors was always synchronous. i.e. when one processor was doing something the other was not. The only exception being interrupts which could be running on the Beeb whilst the second processor was running. The idea behind the 2nd processor was to save on the cost of I/O i.e. Keyboard, mouse, disks etc. It was not really aimed at parallel processing.

Jwoody on 21/9/06 9:16AM
[ Reply | Permalink | Report ]

I correct myself the beeb could get on with vdu commands whilst the second processor was working, but the Iyonix has a graphics processor to do that.

Jwoody on 21/9/06 9:20AM
[ Reply | Permalink | Report ]

There are plenty of things that can be done in parallel, mainly I/O related. A trend amoungst other systems is to have a TCP offload engine. Which is another processor that runs the internet stack taking the load off the main processor of examining incoming packets, allocating memory and performing SSL encryption. All of which can be performed in parallel with the clients of the stack running on the main processor.

druck on 21/9/06 9:36AM
[ Reply | Permalink | Report ]

"There are plenty of things that can be done in parallel, mainly I/O related"

Does not help if the I/O is blocking as most of it is in Risc OS.

Jwoody on 21/9/06 3:25PM
[ Reply | Permalink | Report ]

JWoody: more uninformed comment I see. High level file I/O is blocking, but low level disc I/O and the majority of the other interfaces can benefit from processsor offloading.

If you want to discuss this further take it to csa.hardware as I don't want to see another comment section filled up with your derisorary nonsense.

druck on 21/9/06 5:29PM
[ Reply | Permalink | Report ]

Video playback, block copies (when you're dragging a window around), image resizing maybe even generic tasks like background compiling if a suitably hacky taskwindow-alike module were written.

Trouble really is the wimp is single-threaded, and there is no easy way of changing this without all apps needing changing. The message queue cannot be broken, redraw loops must take priority over everything (unless you're ArtWorks or Java) and messages must be serviced instantly in the given order. But I don't see why command-line apps couldn't be run on a co-processor with sufficient protection put around various SWIs and Clib calls, very much like Java does with its multi-threading.

Piers on 21/9/06 5:58PM
[ Reply | Permalink | Report ]

Druck "JWoody: more uninformed comment I see. High level file I/O is blocking, but low level disc I/O and the majority of the other interfaces can benefit from processsor offloading."

Sorry don't agree. If I make a request to read a file I block. The cpu can get control back to process individual blocks but offloading that is not going to help.

Jwoody on 21/9/06 6:07PM
[ Reply | Permalink | Report ]

FileCore blocks, and would be a nightmare to change. But other FSes would be good candidates to make non-blocking (the Fileswitch API's already there, ISTR). SparkFS would be a prime example, with the decompression being handled in a second thread.

I agree you'd need to modify apps (or ideally the OS) to utilise this though. Without pre-emption, a blocking read will block the system. However, it doesn't block taskwindow, due to an upcall, so with a decent thread module, other threads could continue.

Piers on 21/9/06 6:17PM
[ Reply | Permalink | Report ]

Piers Not sure it would help with SparkFS. If processor 1 makes a request to SparkFS and then has to wait for results its not much use just exercising processor 2. What you would need to do is rewrite sparkfs to be multitasking i.e. Read I/O in one thread and decode in another, trouble is that with current Filecore the I/O read is going to block all threads under that task.

Jwoody on 21/9/06 6:25PM
[ Reply | Permalink | Report ]

We've already done STBs where the Ethernet driver runs on a different ARM core to the rest of RISC OS. And ones where video decoders run on a different ARM core. Fundamentally, the point remains that lots of interesting stuff /could/ be done on the second core. It'd probably be fairly specific stuff to start with so under ordinary use, you wouldn't notice a difference but for those tasks, you would.

7thsoftware on 21/9/06 8:29PM
[ Reply | Permalink | Report ]

"7thsoftware" Are you introducing any parallelism or are you just running on a different core?

Jwoody on 21/9/06 11:11PM
[ Reply | Permalink | Report ]

In the Ethernet instance, there is a main ARM with RISC OS on it and a "front end" part of the Ethernet driver module. Then there's a "back end" part of the Ethernet driver on another ARM core. The back end can worry about some of the buffering and sending stuff to the PHY, plus getting stuff back whilst the rest of the OS can continue with business as normal.

7thsoftware on 21/9/06 11:36PM
[ Reply | Permalink | Report ]

druck wrote>" we wont be able to use the second core for running RISC OS applications, but there is bound to be significant processing tasks that can be off loaded on to it"

Thanks for that Dave, yes of course I simply meant in relation to RISC OS apps (I knew that's what I meant - but didn't write it that way).

AMS on 22/9/06 10:11PM
[ Reply | Permalink | Report ]

AMS: "I simply meant in relation to RISC OS apps"

I'm not sure what you mean here, given the context. Druck said that significant processing tasks can be offloaded to the additional core. Does your reply mean that you don't consider that worthwhile?

Personally, I think that a potential 3x speed increase, accompanied by further speed increases due to some tasks being "offloaded", would be absolutely amazing.

Even more so than the original Iyonix launch, perhaps. (Even the original Iyonix is still much faster and far more capable than any other ARM RISC OS machine!)

Everyone should understand that the release of a dual core ARM most certainly does not mean that a RISC OS machine could be released to take full advantage of it, or anywhere near. But there's no benefit in being unnecessarily negative either.

dgs

dgs on 23/9/06 12:50AM
[ Reply | Permalink | Report ]

dgs>Perhaps I am being unnecessarily obscure (or ham fisted). I was agreeing with Dave Ruck.

dgs wrote>"Druck said that significant processing tasks can be offloaded to the additional core"

Yes and I agreed.

dgs wrote>"Does your reply mean that you don't consider that worthwhile?"

Of course it's worthwhile, I suspect having a 2nd processor present will mean that the inventive developers that are still here will find inventive ways of using this extra capacity. It may be worthwhile that some "standardised" way of interacting with the 2nd processor is instituted. In that way there would be a common API for offloading certain activities to the 2nd CPU and having it "signal" the main CPU when the results of the task were available. That module would be "visible" to the main RISC OS enviroment of CPU 1 and would then trigger the execution, transfer of data to or from, regulate memory use or shared resources of the second processor. CPU2 would have a simplified OS environment and would respond to requests from that other module. It would *not* run a full RISC OS.

I could forsee the any (or all) of the following activities could be performed:

* In memory data processing/indexing/sorting (basically anything that *does not* require access to I/O or require interaction with other tasks) * Precomputing data to be used by the main CPU (it could, for example, create tables of constants for Lookup Table use; do work such as performing Draw matrix transforms, Discrete Cosine transforms). * Offload serial type protocols that require data packaging/unpacking (such as one of the other contributors indicated wrt Ethernet, but why not USB as well).

dgs wrote>"Everyone should understand that the release of a dual core ARM most certainly does not mean that a RISC OS machine could be released to take full advantage of it, or anywhere near. But there's no benefit in being unnecessarily negative either."

If you read one of my earlier contributions you'll find that I said "there is something to be said for using a processor with a large L2 cache, twice the clock speed, support for faster interfaces (like PCI-X/Express) that should give somewhere around x2-x3 times the performance of the Iyonix 80321 even when *only one of the processors is used*. ", I don't read that as negative (nor was it intended to be such). I am saying it's worth doing AS IS. If we get some mileage from the second processor then yep that's better still.

AMS on 23/09/06 6:29PM
[ Reply | Permalink | Report ]

Please login before posting a comment. Use the form on the right to do so or create a free account.

Search the archives

Today's featured article

South East 2006 show videos
Theatre presentations on Drobe TV aka Google Video

35 comments, latest by em2ac on 03/11/06 09:51AM. Published: 30 Oct 2006

Random article

News in brief
Messenger Pro, Wakefield, Ankh, and freeware and shareware developments

31 comments, latest by SimonC on 27/1/06 12:08PM. Published: 26 Jan 2006

Useful links

News and media:
Iconbar • MyRISCOS • ArcSite • RISCOScode • ANS • C.S.A.Announce • Archive • Qercus • RiscWorld • Drag'n'Drop • GAG-News

Top developers:
RISCOS Ltd • RISC OS Open • MW Software • R-Comp • Advantage Six • VirtualAcorn

Dealers:
CJE Micros • APDL • Castle • a4 • X-Ample • Liquid Silicon • Webmonster

Usergroups:
WROCC • RONE • NKACC • IRUG • SASAUG • ROUGOL • RONWUG • MUG • WAUG • GAG • RISCOS.be

Useful:
RISCOS.org.uk • RISCOS.org • RISCOS.info • Filebase • Chris Why's Acorn/RISC OS collection • NetSurf

Non-RISC OS:
The Register • The Inquirer • Apple Insider • BBC News • Sky News • Google News • xkcd • diodesign

Â© 1999-2009 The Drobe Team. Some rights reserved, click here for more information
Powered by MiniDrobeCMS, based on J4U | Statistics
"It does appear that inaccuracy is drobes [sic] house standard at the moment"
Page generated in 0.2636 seconds.