Omega progress clarified | Drobe.co.uk archives

Omega progress clarified

By Chris Williams. Published: 9th Apr 2004, 05:38:13 | Permalink | Printable

Dear Drobe, You're wrong. Love, Dave

Editorial MicroDigital have responded to our recent coverage of the Omega's progress on their website by addressing each of our observations one by one.

Their news release starts by explaining, "We would point out that the Drobe article has raised a number of issues which we hope to clarify here. It is important that anyone reading this article should not take away the idea that we are anti-Drobe or anti Chris Williams, we accept that the article was written in good faith, as is our reply."

We're flattered, truly flattered and it goes without saying that in all seriousness, each and every drobe.co.uk article is checked by at least one other editor for accuracy and editorial balance. We also often consult people who know better than us on a wide range of subjects. Indeed, it is in everyone's interest that drobe.co.uk remains as unbiased, transparent and objective as possible - we can't stress that enough.

MicroDigital have echoed the views of some of our readers on our coverage of the USB issue, by describing our reporting of the matter as "complete nonsense". One particular reader today expressed this opinion via email: "The letter from Simtec is totally consistent with [MicroDigital's] statement. I think you should pull the 'simtec denies' story entirely and include the letter in a footnote to an edited version of the original story."

On the subject of benchmarks, the Yorkshire based hardware developer also illustrated how futile they can be. Oddly, they dismissed Steffen's Iyonix vs. Omega benchmark results on the basis that they were not 'real world' results, despite the fact that Steffen tested both machines in everyday end user situations, such as copying large files with Filer, using SparkFS to make zipfiles and processing JPEGs. Conversely, MicroDigital were quite happy to quote Richard's RISCOSmark results despite them being far from real world tests. And yes, the Omega is included in those results, we're not hallucinating - it's between the Iyonix and SA287 columns. We've also seen the futility of benchmarks, as earlier tonight during investigations, we found that by optimising the RISCOSMark's CPU test, the Iyonix gained a further 30 percent increase in its score while the StrongARM RiscPC gained nothing.

As for the "slam dunk" phrase, we were referring to the Omega's performance in Steffen's benchmarks. Richard's benchmarks highlight that, as far as we understand, the Omega's IDEFS is faster at byte transfers than the Iyonix's ADFS, and that Draw module operations occur faster on an Omega than the Iyonix. The full headache of benchmarking is now apparent, which is why we and the majority of the userbase prefer actual application testing akin to Steffen's measurements as non-technically inclined users can relate better to Photodesk image loading speeds than, say, PCI bus latencies.

Moving along: the discussion on bus bandwidth is rapidly hurtling into the realms of electronic engineering, so we'll leave it up to you to read more about the issue in this Russell King posting on SDRAM speeds, in this guide to memory speeds and also these details on the IOP331, the XScale that MicroDigital have their eye on. Nobody outside MicroDigital can really officially comment more on the exact bandwidth allocations inside the Omega, the design of the shared memory system and how the processors and video FPGA will fit together until either someone suitably dissects an Omega or MicroDigital publish an Omega TRM - we would prefer the latter, naturally.

And so you're probably wondering by now what exactly the point of all of this is? This textual jockeying for position between MicroDigital and the media, the alleged spin versus the supposed propaganda, what will it achieve? For MicroDigital, it gives them a PR boost and more column inches in the news pages and (for some) a welcome return of their Newsdesk section. For drobe.co.uk, it gives us something to write about and investigate and cynically, page impressions to rack up. For end users, it'll hopefully see more facts rise to the surface and nagging issues (such as XScale and USB support) addressed and debated in the public eye - provided of course this whole saga doesn't descend into some tragic soap opera of the kind that this market really doesn't deserve.

And yes, MicroDigital is probably right: it's 'grammatical', not 'grammar'.

Links

MicroDigital

Previous: Simtec denies Omega USB involvement
Next: Adjust ROMs in production

Discussion

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

It can only be a good thing if Drobe are publishing lots of articles and Microdigital are now engaging in debate.

markee174 on 9/4/04 8:13AM
[ Reply | Permalink | Report ]

It's odd that MicroDigital go on about the 80331. It seems that they're trying to justify thaeir claims that there's a great deal more bandwidth to be had from the StrongARM bus - and maybe there is, but the bottom line is that the 80321 is simply a much faster processor, and coupled with the also much faster memory bus in the Iyonix, there's simply no way the Omega can approach how fast the Iyonix - as Steffen has previously pointed out.

As for the 80331 itself, it's completely different to the StrongARM - and indeed, it's very different to the 80321 it's meant to replace. From a commercial viewpoint, the 80331 doesn't even exist (such lead times are common with chip manufacturers), so we're well into the vapourware category.

Certainly, the 80321 does have many bus issues, which the 80331 resolves, but the 321 was most assurdely the right choice at the right time for the Iyonix. Had Castle decided to go down the pluggable processor option, it would have taken much longer, and been rather more expensive, and added to the design complexity. It's easy to demonstrate that there would have been no cost saving for upgraders with this option. As such, it was expected that a new motherboard would be needed should there ever be an Iyonix II, and with that, it would be able to take proper advantages of the totally redesigned bus arrangement in the 331.

As for the questionable claims from MD about benchmarks, I can only repeat what Chris has said - the benchmarks they quote test only a limited subset of activities. As we've already seen, in previous more comprehensive benchmarks, the Iyonix _is_ ~3 times faster than a 200Mhz SARPC in general, and for some operations much more.

What I would like to see is Dave address some of the points I've made. This isn't the first time he's made a general response, and essentially avoided the points I've made.

mrchocky on 9/4/04 9:58AM
[ Reply | Permalink | Report ]

In reply to mrchocky:

"What I would like to see is Dave address some of the points I've made. This isn't the first time he's made a general response, and essentially avoided the points I've made."

MicroDigital's response was to the article Chris wrote, so it seems curious to insinuate that they were 'avoiding' the points that you made (in the Comments?). They may even be reading with the Comments section turned off. I've always found MicroDigital to be very helpful by phone, but if you don't want to ring them, then perhaps you could write an article for Drobe, setting out the issues that interest you. You have a gift for making the technical comprehensible, and that might be to the benefit of all.

Anyway, it's nice to see the issues being debated in a positive fashion ;-)

Stewy on 9/4/04 12:07PM
[ Reply | Permalink | Report ]

In response to the article:

"And yes, the Omega is included in those results, we're not hallucinating - it's between the Iyonix and SA287 columns."

I think they must have been looking at the table of results included in the attachment on the RISCOSmark site -- that *does* omit mention of the Omega.

Anyway, I'm very pleased to see the return of the NewsDesk and the subsequent friendly banter between MicroDigital and Drobe. It's teased out a little more information about the Omega, cleared the air a bit, and provoked interesting debate. It's nice to feel positive ;-)

Stewy on 9/4/04 12:07PM
[ Reply | Permalink | Report ]

If MD put effort in to the product, rather than all this hot air that keeps getting emitted, they might actually release something that folks are able to use and not need to keep trying to defend their unbelievable delays and alleged performance. Just get the bloody thing made and simply /prove/ how fast it is, then we can all be happy. Until then, it's simply meaningless waffle. It's such a shame that Microdigital have turned our platform in to this total fiasco, really. Omega showed potential, oh, several years back, but now..?

For now, I'll have to stick with my somewhat noisy 3.4Ghz PC and VirtualRPC which 'only' emulates a ~450Mhz RiscPC, but does way out perform Iyonix and mythical XScale/Omega in terms of IO, hmm, and cost far less than Iyonix and Omega, *does* run Select, and is real, right now.

Stop waffling. Release good product. Everyone (bar CTL?) happy!

imj on 9/4/04 12:15PM
[ Reply | Permalink | Report ]

The downside being IMJ that not one jot of that hardware was intended for RISC OS thus without the concommitant development of expertise and the associated dependency on part of the computing world I would rather leave alone permanently!

AW on 9/4/04 2:02PM
[ Reply | Permalink | Report ]

imj: From what we've seen so far, the Iyonix would appear to be outselling the Omega by ten or twenty to one. So I doubt Castle would care one way or the other, even if MicroDigital did deliver on some of their promises.

dgs

dgs on 9/4/04 2:04PM
[ Reply | Permalink | Report ]

In reply to dgs:

And from what I've seen, and heard, Virtual RiscPC is outselling the Iyonix. Some, like myself, have purchased the excellent VRPC for use on a laptop, others though are replacing their RiscPCs with the so-called 'hybrid' desktop VRPC machines. Many of those opting for the 'hybrid' VRPC route would appear to be doing so because of worries about their existing software being incompatible with the Iyonix -- in many cases, a simple upgrade would resolve that issue, but it's the *perceived* worry that matters.

MicroDigital, of course, have kept the most significant of their pledges, which was to build a machine which was totally compatible with existing software. Once the remaining features are enabled, the Omega may well help to keep the 'native' RISC OS market alive. As you said, it's choice that's important ;-)

"So I doubt Castle would care one way or the other, even if MicroDigital did deliver on some of their promises."

I'm glad to see that you're not one of those who argue that the Omega is 'depriving' Castle of potential Iyonix sales.

Perhaps Castle don't care. But why do *you* care so much? As you pointed out, I've purchased an Omega, so things like USB matter to me. You've got your Iyonix. You've often stated how pleased you are with it. Why all the effort expended upon criticising MicroDigital in public?

Stewy on 9/4/04 3:26PM
[ Reply | Permalink | Report ]

dgs> You may have missed my point. Microdigital appear to be claiming that their mythical XScale Omega would outperform Iyonix. *If* they ever make it (unlikely, IMO) and *If* it's better than Iyonix (again, I can't see how) then that would give Castle mild cause for concern. The statement was a loaded one. ;-)

imj on 9/4/04 4:01PM
[ Reply | Permalink | Report ]

IN GENERAL

RISC OS 5 is not that much of a step.

these "hybrid" machines are complete and utter cr*p

Any currently outstanding compatibility issues are generally sorted with aemulor.

Iyonix costs what it does because it is a *very* niche machine for a small market.

If you cannot afford one then stick to the pants that RO4 is becoming

enjoy...

epistaxsis@work on 9/4/04 6:15PM
[ Reply | Permalink | Report ]

epistaxsis: Unfortunately some of us _have_ to buy Windows machines because of work or some software they have to have. Would you rather they bought a 'complete and utter cr*p' hybrid machine, and kept using (and hence) supporting RISC OS software, or were forced to leave the market entirely?

Your all-guns-blazing, VRPC-bashing attitude seems somewhat ignorant of many of the issues involved, to me.

hutchies on 9/4/04 7:01PM
[ Reply | Permalink | Report ]

Having just skim-read the Microdigital article it seems an extremely pleasant turnaround from their previous, very hostile-seeming (even if not intentionally) attitude towards the press. Long may it continue!

hutchies on 9/4/04 7:03PM
[ Reply | Permalink | Report ]

Stewy: Good question. I guess it comes down to the fact that I don't like people being misled. (Though, as others have pointed out, you'd have to be pretty gullible to *still* take it all at face value).

dgs

dgs on 9/4/04 7:08PM
[ Reply | Permalink | Report ]

I'm pleased to see an intelligent level of debate around Omega. I happen to believe that competition amongst the dedicated (ie ARM-based) RISC OS computers is a good thing.

The only caution I would make is that we should not compare released with unreleased computers. Iyonix is well over a year old now while Omega/X-Scale is not yet available. It is incontrovertible (and unsurprising) that Iyonix is faster (in a general usability sense) than Omega/SA. When Omega/X-Scale is released, it will be a (much?) newer computer so we would hope that it would be faster - time will tell - but Iyonix is currently the faster available machine.

I just hope that Omega/X-Scale *is* faster... and I hope that it provokes Castle into a faster Mk2 Iyonix. THAT'S what competition is supposed to achieve.

TonyStill on 9/4/04 7:20PM
[ Reply | Permalink | Report ]

I think many people will be waiting to see if USB also materialises for the Mico

Sound was fixed. Networking is available. (Joystick support was provided by a 3rd party)

USB is still outstanding.

jess on 10/4/04 10:59AM
[ Reply | Permalink | Report ]

Networking is available for the Mico? Have you actually got it? Cheers!

The Doctor on 10/4/04 5:14PM
[ Reply | Permalink | Report ]

No, (my brother doesn't need it at present), but I've exchanged messages with someone who has.

jess on 10/4/04 6:12PM
[ Reply | Permalink | Report ]

TonyStill> There is *no* reason to suppose there'd be much speed advantage to an Omega with the xScale card fitted. After all it's *not* a whole new computer, all that's changed is the processor. The southbridge (and its harddisk controller) will remain the same (so no speed increase there then), yes it can be argued that the Northbridge *can* be reprogrammed - but as the RAM sockets are soldered in you're not simply going to be able to reprogram it to use *newer* RAM and get much of a boost there either (IMHO).

Where it *would* score is it would fulfill a *bullet point* feature that many people paid for nearly 4 years ago, it would probably offer some speed improvement over the existing Omega processor arrangement - but here's the catch - what impact will "ArmTwister" have (I can't really believe that it will have zero impact). Remember much of the OS is 26bit - so therefore that will either have to be "ArmTwistered" so it can run on the xScale (or more likely) it may just run on the *existing* StrongARM (therefore *No speed improvement*) while 32 bit code would be faster - as they could be run on the faster xScale but the moment they need to access the OS there'd be a speed hit (as it would have ArmTwisters overhead and/or would need to be run on the current StrongARM).

As to the assertion that's floating around here *No* Omega is *not* a newer computer than Iyonix its RAM, Processor are older to cap that the processor it uses is out of production (hence obsolete) simply plonking in a new processor does not make it a *newer* computer than Iyonix - I'd also point out that the FPGAs MD uses although reprogrammable are themselves old (by todays standards) and simply reprogramming them won't change that will it ?

The fastest xScales are at the 733MHz mark, Iyonix runs at 600MHz. For purely register based (no I/O) the speed improvement would probably be commensurate with the difference in clock rate (at best 22%). Once I/O gets involved some of that has to be piped through the OS which is still 26 bit and which *can't* run natively on the xScale (it follows that it will be *slower*). So any speed increase will depend on the proportion of code that is 26 bit and the proportion that is 32bit. 32bit code that makes no I/O or OS calls will be quickest, 32bit code that invokes 26bit OS/ or I/O code will be slower. Yes no one can be certain of what speed increase, if any, is likely but it can be *no faster* than 22% in Omegas favour - and may well be either less than this or even less than the current performance of the Iyonix (depending on the precise code run and the effectiveness of ArmTwister).

As to the notion that a faster Omega will *provoke* Castle into doing an Iyonix-II, the fact that Iyonix is *already* faster and is already here does not seem to have make Omega die-hards move over to Iyonix... if a putative IyonixII existed even if it was ten times faster than Omega some people here would still opt for Omega - Castle may well have won the intellectual case but for those who have "faith" in Omega - mere logic doesnt seem (sadly) to carry much weight.

Regards

Annraoi

AMS on 11/4/04 12:49PM
[ Reply | Permalink | Report ]

AMS: My main point was not to confuse what we have now with what we may have in the future (and that's a faster Iyonix). Frankly, I can't see that Omega/X-Scale is going to be faster than Iyonix either but I'd rather give them the benefit of the doubt.

As to a Mk2 Iyonix, it's me that wants that (in a year or two, Mk1 Iyonix is still exciting enough for me for now), I don't mind the Omega fans not buying it. And I think it's Virtual Acorn that will be the performance threat, BTW.

TonyStill on 11/4/04 7:22PM
[ Reply | Permalink | Report ]

Let's all be realistic and accept that a complete replacement for the Iyonix is probably 3 to 4 years away, if ever.

dgs

dgs on 12/4/04 2:26AM
[ Reply | Permalink | Report ]

As for the benchmarks and MDs dismissal of Steffen's Iyonix vs. Omega but liking of Richard's RISCOSmark:

Steffen did indeed run tests with normal end-user tasks. whereas the RISCOSmark measures the odd system feature's performance. To put it simple: I really do *not* care at all how fast memory or harddisc transfers run in low level - all I care about is the speed and stability I get in real use of the system. If the system performance is achieved by good programming of the hardware drivers or by using fast hardware - what counts is what good it does me.

For this reason benchmarks like the one from Steffen tell me what I want to know whereas RISCOSmark doesn't. But I can understand MD preferring RISCOSmark ;-)

hzn on 12/4/04 9:06AM
[ Reply | Permalink | Report ]

Although somewhat off-topic, I was struck by the very small advantage shown in Steffen's benchmarks by the overclocked S/Arm RPCs compared to the 202Mhz ditto - between 0-10%. Bearing in mind the Kinetic was demonstrated to be about 50% faster than the latter in a series of real-world tests done for Eureka when it came out, I suspect there mightn't be much to choose between a 300Mhz Kinetic and the Omega....

bucksboy on 12/4/04 7:02PM
[ Reply | Permalink | Report ]

It is possibly an unfortunate selection of benchmarks that suggests that overclocking won't gain much. There is no real "CPU bound" benchmark in my selection. The one that comes nearest is the SparkFS compression benchmark, but this still also depends significantly on I/O performance (disc and memory).

Originally, I "designed" the PackDir benchmark to be a CPU benchmark (hence using the RAMDisc to minimize I/O impact), but as you can see, it is nearly totally memory speed bound. Try it on a Mico or a RiscStation and see them outperforming your Risc PC significantly...

As I said on my benchmark page, if someone has a suggestion for other real-world benchmarks, drop me an email...

hubersn on 15/4/04 2:02PM
[ Reply | Permalink | Report ]

As for "dismissed Steffen's Iyonix vs. Omega benchmark results on the basis that they were not 'real world' results" ... taking a closer look at the NewsDesk it says "Regarding Steffen’s test results we should point out that we have not said that they were in anyway a scientifically accurate comparison between the machines, nor do we accept that his results are necessarily representative of the real world. Whether any particular benchmark is relevant or useful depends entirely on what you want to do with your computer. That said, we again thank Steffen as his results were helpful to us."

I think that that is a bit different - they basically said that the tests are not scientifically accurate (what test is, as a matter of fact?) and that they're not representative of the real world. This can e.g. mean that Steffens tests missed the odd type of work you can do with such a system. One that e.g. comes to my mind is the time the computer takes to fire up and pick up the new email.

So suggestions are welcome for further tests as Steffen did ask for above and on his website.

hzn on 27/4/04 3:27PM
[ 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

NetSurf bags GBP10K investment from Google
Four uni coders lined up to improve web browser

35 comments, latest by tlsa on 24/11/08 4:19PM. Published: 21 Apr 2008

Random article

RPCEmu Spoon Edition 0.8.2 released
This flavour of the open source RiscPC emulator includes various fixes and can also be built for Apple's Mac OS X computers.

Discuss this. Published: 23 Feb 2009

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
"Most [articles] are unhealthily subjective, biased and increasingly boring rambles from the same few writers. Drobe is a tabloid. And its writers certainly not 'press'"
Page generated in 0.1037 seconds.