Omega XScale support in 2004? | Drobe.co.uk archives

Omega XScale support in 2004?

By Chris Williams. Published: 20th May 2004, 22:34:12 | Permalink | Printable

Didn't we ask this last year?

In the latest newsletter from Desk, published yesterday, the Dutch dealer has reported that Microdigital are currently focused on completing their 'ARMTwister' technology for the StrongARM powered Omega computer. Desk have suggested that Microdigital are aiming to have demonstrable, working XScale support ready in time for the Expo show.

Desk are also the main sponsors of this year's Expo event, which is to be held next month in the Netherlands.

The Omega currently runs RISC OS 4, which is a 26bit OS, whereas the Intel XScale processor that MD want to use is fully 32bit and has no 26bit support. According to Microdigital, the ARMTwister is "a JIT (just-in-time) code translation system to convert 26bit-only instructions into the 32bit equivalent ones, on the fly. It is implemented in the Omega NorthBridge FPGA. This takes advantage of the natural parallelism of hardware to implement the various table look-ups and calculations necessary to convert ARM code, as it is loaded into the CPU instruction cache."

The Omega was first announced in October 2000, and eventually shipped in 2003.

Links

Desk website RISC OS and 32bit, from riscos.info

Previous: Castle opens RISC OS future wishlist
Next: Wakefield 2004 theatre talks

Discussion

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

I'm betting that they'll complete it Just In Time for the show...

(OK, so maybe I'm not that optimistic, but that pun was too good to pass up!)

Phlamethrower on 21/5/04 3:13AM
[ Reply | Permalink | Report ]

hmm, might've missed the boat already, there is a better option available... [link]

lostamarble on 21/5/04 3:30AM
[ Reply | Permalink | Report ]

I'm betting that ARMtwister will turn out to be a 32bit RISC OS running 32 bit software on the Xscale. The StrongARM will be disabled, with a promise that a future software version will use it to run 26bit tasks.

cbcbcb on 21/5/04 8:00AM
[ Reply | Permalink | Report ]

In reply to lostamarble Thanks for that link - very interesting. So, 2.6 GHz RISC machine sometime mid-2006 perhaps ?

In the meantime, I'm as curious as everyone else to see what ARMTwister amounts to. Good luck with bringing it to market, MicroDigital.

martin on 21/5/04 8:00AM
[ Reply | Permalink | Report ]

I suspect that ARMTwister will be a hardware base equivalent of Aemulor, that will be hardwired into every Omega.

JWCR on 21/5/04 9:42AM
[ Reply | Permalink | Report ]

Wouldn't it be more sane to get the basic machine stable first, rather than faffing about with features that'll only stress it more.

imj on 21/5/04 10:26AM
[ Reply | Permalink | Report ]

Apparently from what I heard the machine is now meant to be 'stable'. Which means they can now start faffing about with features that'll stress it more.

piemmm on 21/5/04 10:52AM
[ Reply | Permalink | Report ]

The one ROL had at the show was pretty knackered. No doubt MD will use the nice get-out-clause "firmware upgrade" YET again, but that windowing bug, for example, has been there for YEARS now (literally) so I don't expect they'll be fixing it any time soon. The machine kept hanging and totally locked up several times in just 10mins of gently prodding it. If MD can't even work with their OS supplier and get them a stable machine, what hope do users have? Alan Gibson claimed his Omega was stable throughout the show, but then again, it wasn't a production machine. A saving grace is that disk I/O seemed very fast - the replay movie which ROL showed running full rate apparently only runs 2-3fps on a RiscPC. IMO, the problem with the RiscPC isn't the CPU really, it's the bottlenecked I/O system - so a stable Omega would certainly be very useful, even without XScale / thingytwister. A stable Omega... it's almost an oxymoron.

imj on 21/5/04 11:31AM
[ Reply | Permalink | Report ]

imj: Wouldn't it be better if Microdigital accepted the inevitable and licenced RISC OS5 of Castle. After all, that has proven stable for over 18 months.

JWCR on 21/5/04 11:58AM
[ Reply | Permalink | Report ]

And with the 26bit processor no longer available, they will have to move exclusively to XScale eventually. Why not make it sooner rather than later.

JWCR on 21/5/04 12:00PM
[ Reply | Permalink | Report ]

cbcbcb: Or what about a 32 bit RISC OS which can run RO 4 in a window (like a risc pc and pc card)?

jess on 21/5/04 12:46PM
[ Reply | Permalink | Report ]

JWCR> Youwhat? What's that got to do with making a stable machine? And you really think Castle would have any interest in helping their competitor out with an OS for their box? With MD's continued claims that Omega would somehow be faster than Iyonix, it'd seem wise for Castle to keep their OS to themselves.

Anyhow, AIUI, ROL's offering is prettymuch 32bit ready - just needs a customer. I'm not banging the old drum again, but some of us prefer a richer flavour of OS. ;-)

What "the inevitable" referrs to, I've no idea.

imj on 21/5/04 12:47PM
[ Reply | Permalink | Report ]

imj: I was at Wakefield and agree that the Omega running Adjust on the RISC OS Ltd stand was very unstable, although in a different way to the RiscPC that was behaving unpredictably on Paul Middleton during his presentation on Saturday morning. I can only attribute this to either Adjust or RISC OS Ltd, as after using an Omega for many hundreds of hours, I've found it to be rock solid (I've only ever run RISC OS 4 though.) To be honest, from both my personal experience and the people with Omegas who I know, there are very few who have any problems at all. The windowing bug you speak of as having being present for years has never happened to me, even before any firmware upgrades. In fact, I'd be pushed to tell of any bug that I've noticed other than that Doom has the wrong palette (I'm presuming this is a VIDC emulation issue).

not_ginger_matt on 21/5/04 1:02PM
[ Reply | Permalink | Report ]

imj: "And you really think Castle would have any interest in helping their competitor out with an OS for their box?" Castle would be delighted if any company wanted to put out a computer running RISC OS5 , as they would make some money from the licenced copy of the OS on those machines. It is probably true that they would prefer if everyone bought Iyonix variants, but as they say, "a bird in the hand...".

JWCR on 21/5/04 1:44PM
[ Reply | Permalink | Report ]

In reply to imj:

As not_ginger_matt has already noted, the Omega is very stable. My own Omega hasn't ever manifested the sort of problems that blighted the machine on the ROL stand. In contrast, I can confirm that the Omega on Alan Gibson's stand had none of those difficulties. As far as I'm aware, Alan Gibson's Omega *is* a production machine, just rehoused in a gorgeous, slimline glass rack ;-)

With networking now shipping for the Omega, it's exciting to hear that the XScale may be ready for the Dutch Expo Show.

Stewy on 21/5/04 2:02PM
[ Reply | Permalink | Report ]

I suspect we may see an Omega with an XScale card in, although I very much doubt we'll see one without a StrongARM in alongside.

Of course, I will be pleased to be proved wrong!

spellinn on 21/5/04 2:15PM
[ Reply | Permalink | Report ]

What's wrong with having a StrongARM in the machine, too? It won't do any harm, it will simply be disabled.

And I don't think that we will see XScale cards on the Expo, but development mainboards with both StrongARM and XScale on board, to show that the ARMTwister design does work. The finished XScale cards for production Omegas will probably take a few weeks more (for production, packaging, installation manuals, delivery, etc).

JGZimmerle on 21/5/04 3:56PM
[ Reply | Permalink | Report ]

IMJ:

"Anyhow, AIUI, ROL's offering is prettymuch 32bit ready"

Their OS? Really? Wow. I never thought we'd see a 32bit version of ROL's flavour of RISC OS

-- Spriteman

Spriteman on 21/5/04 4:32PM
[ Reply | Permalink | Report ]

Spritey> Well, ROL have been saying that they'd do a 32bit version since day 1 of Select. Paul Middleton has reiterated this several times that I've seen reported. I'm told lots of Select can already be built 32bit-happy, it just needs a customer. I'm sure everyone's quite aware why Iyonix doesn't run a RISCOS Ltd OS - ROL simply weren't asked - but maybe Microdigital will need a 32bit RISC OS. There's not much point in ROL doing a whole pile of work to produce a 32bit RISC OS for RiscPCs!

imj on 21/5/04 7:44PM
[ Reply | Permalink | Report ]

imj: My Omega is a production machine (Issue 1 board). It's just in a rack case so has a different PSU and CD drive. The screen redraw has TTBOMK only appeared when using Select; I've never seen it under 4.03.

liquid on 21/5/04 9:57PM
[ Reply | Permalink | Report ]

liquid: So some Omega bits work reliably under 4.0x, and some bits work reliably under 4.3x ?

Errrr, this doesn't sound promising

All that talk about splits in RISC OS, and now MicroDigital are having an OS split even on the same hardware...

dgs

dgs on 21/5/04 11:48PM
[ Reply | Permalink | Report ]

I use 4.37 on my Omega all the time and have not experienced any graphics problems.

JGZimmerle on 22/5/04 8:46AM
[ Reply | Permalink | Report ]

All hardware/OS/application combinations have odd problems for some people its the nature of the complexity inherent in the interactions between the three. Although it seems OK to slate Omegas for this as if its a global problem and god forbid if Iyonix issues are reported (I can feel knives in the back already).

I have no personal experience of Omega,Select or Adjust but it is clear to me that the following is taking place:

1. Omega is as stable as an Iyonix. 2. Select/Adjust OS is or nearly is 32 bit compliant. 3. The Xscale for Omega is winding up for production.

regards,

Malcolm

mripley on 22/5/04 10:22AM
[ Reply | Permalink | Report ]

Malcolm>

Omega is (basically) a RISC PC with FPGA's stepping valiantly in for the IOMD and VIDC. The biggest change is in the RAM and the Southbridge. Except for HDD access there's little to choose between it and (say) a Kinetic.

I really like your comment that Omega is as stable as an Iyonix - but then state before hand that you had no personal experience of Omega, Select or Adjust - so how exactly did you arrive at comment 1 (above) ?

1. Omega is as stable as Iyonix ?

Well I can't really confirm that can I, as MD have *never* chosen to exhibit any any of the three RISC OS shows I've attended since 2000. If it is *really* as stable as is claimed I am quite happy to accept what I see with my own two eyes. If MD can't be bothered then I'll have to contend that above statement is unproven (and as you haven't seen it either you may wish to modify your above statement surely).

2. Select/Adjust OS is or nearly is 32 bit compliant

All it takes is *one* instruction in the right place for it to blow up so being "nearly" 32bit is meaningless - it's one of those cases where the thing either IS or ISN'T (this is computing things are TRUE/FALSE and not SORT OF....)

3. XScale for Omega is winding up for production.

I'll believe it when I see it. Which version of the xScale is it by the way ? The one MD were originally going to use has gone out of production during the long 4 year wait.....

Kind Regards

Annraoi

AMS on 22/5/04 1:07PM
[ Reply | Permalink | Report ]

AMS: "how exactly did you [Malcolm] arrive at comment 1 (above) ?"

He's presumably saying it on the basis of the two or three people who have claimed that their Omegas are totally stable.

But alongside that "evidence", there's also the two or three Omegas we know of that have died.

So, out of a pretty small total number of Omegas that have shipped anyway, all we can conclude is that some have died, and some haven't died.

dgs

dgs on 22/5/04 1:26PM
[ Reply | Permalink | Report ]

Wich Omegas have died, then? And have they been returned to MD for repair? I once managed to crash mine while updating the firmware, so I sent it in and had it back within the week, and that was from Germany to England and back to Germany.

JGZimmerle on 22/5/04 2:23PM
[ Reply | Permalink | Report ]

JGZimmerle: Two that I know of died of their own accord, and had to be sent back to MicroDigital (couldn't be fixed even by the local MicroDigital dealer). I think I had heard about a third one that suffered the same fate, I'm not sure of that one. Plus there's the one RISCOS Ltd have, which seems famously unhealthy. Perhaps there are more that we don't know about.

It's good to hear that (from your experience) turnaround times for repair are pretty quick.

dgs

dgs on 22/5/04 2:32PM
[ Reply | Permalink | Report ]

@dgs:

Let me be the fourth person to say this: "My Omega is stable"

Regards,

Rick

rdenk14 on 22/5/04 3:42PM
[ Reply | Permalink | Report ]

With the Omega now starting to become a viable computer, perhaps I might purchase one this year. The important bit for me was networking. And now it reportedly works.

sa110 on 22/5/04 4:28PM
[ Reply | Permalink | Report ]

So can Omegas recognise 1Gb of RAM as advertised on the MD website yet? If not then what is the current maximum?

blahsnr on 22/5/04 7:29PM
[ Reply | Permalink | Report ]

blahsnr: 1GB RAM is (apparently) currently quite difficult to source for the Iyonix, as well.

dgs

dgs on 22/05/04 8:33PM
[ Reply | Permalink | Report ]

Really? 1 GB DDR266 DIMMs seem to be on sale all over Germany - Infineon, Kingston, Corsair...around 270 EUR. Why shouldn't those work?

hubersn on 22/05/04 9:33PM
[ Reply | Permalink | Report ]

hubersn: Castle haven't tested one recently, and hence aren't currently selling them. They have agreed to investigate this and come up with a conclusion.

It may also be solved from another direction, as the hybrid project in Australia plan 1GB RAM for several of their early machines. Once a specific part is tested and known to work, of course anyone can buy it and fit it themselves, or ask their local dealer or Castle to fit it.

For those wondering why anyone would need more than 512MB RAM, I would quite like it for Photodesk, and for Linux. Admittedly, either will run more than happily in 512MB anyway.

Any news on Linux for the Omega, anyone?

dgs

dgs on 22/05/04 10:08PM
[ Reply | Permalink | Report ]

Oh yeah, I should add that it was entirely my own fault when I crashed my Omega while flashing the firmware, I had not followed the recommended procedure.

JGZimmerle on 23/05/04 3:46PM
[ Reply | Permalink | Report ]

OK, as a mere user of RISC OS hard- and software (SARPC with 287Turbo and IDEFS plus Adjust), and therefore representative of the wider market should RISC OS ever resurface as a mainstream OS, it strikes me that Adjust is a beautifully logical and user-friendly version of the OS, and much more feature-rich than RO5. (If you take the comparison of a car driver with a mechanic, I am a driver, so what happens 'under the hood' is not a major concern of mine, as it would not be with a mainstrean user.)

IMHO, an Omega running RO Adjust on either 26 bit with ArmTwister or on 32 bit (should that ever happen), both with a 1.2GHz XScale, will make an Iyonix look a little.... dated... and ... slow... so I am saving my money in the hope that Omega will beat the critics and deliver.

That said, if Castle produce Iyonix2 with a CPU socket that allows a processor upgrade, everyone would be interested. However, I reckon that the vision lies with a version of RO that keeps evolving to match the advantages of XP, Longhorn or whatever. And as a user who has to cope with Windoze at work, the MAJOR advantage of RO is the drag-and-drop saving. My network Windoze defaults to my root directory whenever I want to save, even if I have already saved to a specific directory, which is both inefficient and irritating.

So, until Omega has XScale or Castle produce Iyonix2, my money will stay unspent and my RPC will soldier on. Blame Acorn for building things to last.

BigMac on 23/05/04 5:58PM
[ Reply | Permalink | Report ]

@dgs:

Yeah, you said you knew of a few machines wich failed and I asked for names. If you don't want to tell them publically, feel free to mail me.

I guess there can always be a few faulty machines in a production run, however MicroDigital never had to call back all machines of a production run to fix a design fault.

JGZimmerle on 23/05/04 9:27PM
[ Reply | Permalink | Report ]

jgz No MD don't recall machines.

When their ethernet driver fails to work properly with the base OS they ship the Omega with (and the Mico come to think of it), they just tell you that you have to upgrade your OS if you want reliable networking.

Now that really is great customer service isn't ;o)

blahsnr on 23/05/04 10:01PM
[ Reply | Permalink | Report ]

JGZimmerle: I'm not sure exactly what you mean by your last comment.

dgs

dgs on 23/05/04 10:11PM
[ Reply | Permalink | Report ]

This is not entirely true. You can have reliable networking with the base OS, too. You just have to choose the right networking software. And if you choose to upgrade to Select or Adjust, you get quite a bit of extra-value, apart from the fixes in the networking software. Compare this with quite a hefty postage fee (at least from Germany) or a costly service fee from STD, just to get the design-fault of your Iyonix motherboard corrected.

JGZimmerle on 23/05/04 10:26PM
[ Reply | Permalink | Report ]

jgz If you have to choose the 'right networking software' then how can you claim that an Omega is compatible with all software?

That was the basis of using the obsolete StrongARM processor and RISC OS 4 making sure it would be compatible with all 26bit software. Now it obviously is NOT compatible with all software.

As the Mico suffers from the same ethernet problems whereas the very similar performing RiscStation doesn't the fault must lie with the driver.

blahsnr on 23/05/04 10:33PM
[ Reply | Permalink | Report ]

JGZimmerle: My Iyonix doesn't have a "design-fault motherboard".

What on earth are you wittering about? Is this more of your "marketing-nonsense" that you talked about previously?

dgs

dgs on 23/05/04 10:46PM
[ Reply | Permalink | Report ]

IIRC the Omega was meant to be compatible with just about all the Software wich was StrongARM and ROS4 compatible. This still holds true, because AFAIK the networking software that does not work is failing because of some bugs in the internet stack of ROS4.

JGZimmerle on 23/05/04 10:49PM
[ Reply | Permalink | Report ]

@dgs: Maybe yours does not, but a lot (IIRC the complete first production run) had. See [link] for details. And IIRC the "solution" was to disable the 64-bit mode of the PCI bus, effectively halfing the available throughput.

JGZimmerle on 23/05/04 10:58PM
[ Reply | Permalink | Report ]

jgz The Omega is EITHER compatible with ALL software that runs on a StrongARM and RO4 OR it IS NOT.

Compatibility is like pregnancy, you cannot be 'a bit pregenant' and you cannot be 'a bit compatible' ;o)

Personally on the three RiscPCs I have installed RISC OS 4 on I have never experienced networking problems and two of them were StrongARM machines.

blahsnr on 23/05/04 10:58PM
[ Reply | Permalink | Report ]

jgz I will treasure your last comment to the end of my days ;o)

I am off to bed now but thanks for the laugh, you have just cheered me up no end :o)

Don't forget that Castle have acknowledged the problems publically, explained the situation to the press and fixed these issues free of charge for those who wanted the fix.

Contrast this with Microdigital know there is something wrong with their ethernet driver, but instead of fixing it, they come out with some pseudo technical excuse, and then tell you you have to upgrade your OS. Which if you don't already have Select will cost you a hefty ukp155.

Schlaf gut, bis morgen schatz!

blahsnr on 23/05/04 11:08PM
[ Reply | Permalink | Report ]

@blahsnr:

I am not your "schatz"!

Well, Castle may do the fix for free, but will they also take over all the shipping costs? As I said, it is quite expensive to send a complete computer from Germany to England, and I presume it isn't very cheap inside the UK, either. Of course you could take it to a show they attend, but then you would have to pay for the travel as well. Of course you get other benefits when going to a show in addition to the fix for your computer, just like with buying Select for an Omega. And Castle have not really fixed the problem, they are disabling a part of the Iyonix's functionality to make the machine stable. IMHO this represents a loss of value.

And compatability is not like pregnancy. If almost all software that runs on a StrongARM+ROS4 RiscPC also runs on the Omega, then it is mostly compatible. No one could guarantee that any machine would be 100% compatible. In fact not even all StrongARM-RiscPC are 100% compatible with each other, there have been different motherboard versions.

Since the networking software you use on your StrongARM+ROS4 RiscPC works fine there, it most likely will work fine on the Omega, too.

JGZimmerle on 23/05/04 11:27PM
[ Reply | Permalink | Report ]

You guys seem to be very upset with each other, but really it just looks like you need to calm down a little. If JGZimmerle has problems with his Iyonix, then I'm sure Castle will be happy to talk to him about the easiest ways to resolve the alleged problems.

I'm totally lost when reading the claim that any networking code that works OK on the Risc PC "most likely" will work OK on the Omega. Surely it is for MicroDigital to test out the code, then present it fully working at shows or user groups. If it doesn't work, then MicroDigital need to fix it. Why has it taken so long?!?

Presenting code or hardware and saying "it is working" is pretty meaningless, if the company can't also prove that in real life. Is it true that the MicroDigital "test bed" at the Utrecht show will only be a development board, not an actual X-Scale card working in an Omega?

Simtec demonstrated a RISC OS laptop fully working at a Wakefield show quite a few years ago, remember...

dgs

dgs on 23/05/04 11:46PM
[ Reply | Permalink | Report ]

dgs I'm doing fine, Julian seems a bit miffed that I called him the equivalent of 'darling' ;o)

blahsnr on 24/05/04 09:40AM
[ Reply | Permalink | Report ]

@Stan:

Let me get things straight: There is nothing wrong with the driver that MD wrote.

There is a problem with the networking stack in Risc OS. This problem didnt showed up earlier since the risc pc's do not operate on the same speed as the Omega does. And the fact that there is a SA in the Omega and in your RPC's does not make them the same computer. A man with your qualifications should know better ;-]

Both Md and ROL are not to blame for this because that particular piece of the internet stack was written before ROL started working on Risc OS.

So youre conclusion on that there is something wrong with the driver is very "kort door de bocht"

Pravda, Tawarisj !

regards,

Rick

rdenk14 on 24/05/04 10:09PM
[ 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

Running RISC OS Programs on Linux
Peter Naulls contemplates the possibilities

17 comments, latest by drjones69 on 6/6/05 11:17AM. Published: 31 Mar 2005

Random article

Iyonix Select yet to show 'commercial viability'
Pledged Select cheques won't be cashed until work begins

30 comments, latest by jess on 24/6/05 3:24PM. Published: 23 Jun 2005

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

© 1999-2009 The Drobe Team. Some rights reserved, click here for more information
Powered by MiniDrobeCMS, based on J4U | Statistics
"Please give the Castle bashing a break... please!"
Page generated in 0.2813 seconds.