Omega production saga continues - RISC OS News, Software and Information

Omega production saga continues

Published: 18th Feb 2004, 15:14:20GMT Source: drobe.co.uk
By Chris Williams

Page 1 of 1

Photos, excuses, hopes

Last week we reported that Liquid Silicon were in the process of clearing their Omega order backlog. This week Dutch dealer Desk have announced that they too are assembling their share of MicroDigital's StrongARM powered computers.

Desk have also uploaded photos of their Omegas online, revealing one or two interesting things. Firstly, it would appear that MicroDigital have produced or commissioned a support CD for their users. Secondly, the Omega motherboard is copyrighted 2001 to 2003 and labelled as 'issue 2'. Also on the PCB silk screen, you can clearly make out "with ARMTwister Technology(tm)".

Omega issue 2 motherboard

The motherboards pictured boast a StrongARM processor hardwired to the PCB, an Xilix FPGA and an ALi chipset. We can also make out 4 PCI slots, a floppy drive socket, 2 IDE sockets, 2 RAM banks sockets, some Flash memory, sockets for the RISC OS 4 ROM set, a huge heatsink for the graphics FPGA ('Lightning'), some other glue electronics and the standard I/O ports. There's also a sticker showing copy '300060' of RISC OS 4. The cynical among us will presume the 3 refers to the serial number subset allocated to MicroDigital and that 60 is the unit number for that ROM. Go ahead, use this to guess how many Omega sales there have been, we know you're dying to.

In a newsletter this week, Desk told its customers: "With regards to the Omega, we don't want to keep people in anticipation any longer. While we have been waiting for a breakthrough to happen, we have wisely decided to keep quiet with respect of the Omega."

On the gradual distribution of the Omega, Desk added: "The userbase is growing steadily. On the club circuit the Omega will become more and more visible."

Desk, a MicroDigital partner, also explained that the delays in producing the Omega were down to testing and fixing completed motherboards. According to Desk, for every batch of manufactured Omega motherboards, "about half" need tweaking and fixing in order to work. Unlike a larger manufacturer, MicroDigital can't afford to order a pre-production batch to test and improve their product, before ordering a run of, say, 100,000 motherboard units. Economies of scale limit them to ordering small runs of boards, with MicroDigital's designer Dave Prosser concentrating on getting boards working rather than addressing missing features.

Desk doesn't appear to be too happy with the situation, commenting: "It is a pity that MD have not found a solution to this problem, such as employing someone to accelerate the rate of progress on all other unfinished aspects of the Omega such as software for the ethernet card."

Ah yes, the ethernet card. One of the promised features that is still not complete. As far as we're aware, ethernet, USB and the promised Lightning graphics system are missing and still in development. Apparently, the design of RISC OS limits the effective speed at which data from a network card is read, which is true to an extent. RISC OS reads packets from network cards on the centisecond timer, which isn't fast enough for 100MBit or 1GBit network cards. Presumably though, something could be worked out between the hardware and the current OS developers. Anyway, this flaw in the design of RISC OS is blamed for the delay in releasing an ethernet driver for the Omega.

History lesson
The Omega was first touted in October 2000 with some quarters of the RISC OS media expecting it to arrive at the start of 2001. The Omega was full of hope and ambitious promises of XScale co-processors and its own graphics engine. The public first saw the Omega in May 2001 at the Wakefield show, as MicroDigital demonstrated a reportedly unstable prototype. In September 2001, MicroDigital's Dave Atkins told Archive magazine that his company was aiming for a release in October 2001 and that the Omega had till then "cost double" their budget.

MicroDigital next appeared in public in June 2002, at the RISC OS Expo. This time they had a more or less working prototype, which was tied to a PC to kick start it into booting. The Omega then went on a usergroup and press tour before the west Yorkshire based hardware company promised to have the Omega delivered by Christmas 2002 and the ARMTwister magic technology would be ready by January 2003. After busting through yet another deadline with no goods on users' doorsteps, MicroDigital had no explanation to offer. Dave Atkins briefly appeared at the February South West show with a few Omegas for selected, privileged developers and dealers.

Despite the delays and missed deadlines, MicroDigital sent letters to Omega deposit holders in March 2003 asking for full payment because the Omega was apparently within days of completion and delivery. MicroDigital were around the time also grumbling about RISC OS licensing issues and precisely who owned what. In May 2003, at the Wakefield show, MicroDigital demonstrated more working Omega computers, although they were lacking support for USB, floppy drive, hardware JPEG and MPEG handling, graphics acceleration, SCSI, networking and the all important XScale. In June 2003, a few machines were handed over to selected users at the RISC OS Expo 2003. show. In August 2003, we heard from our first reader who had finally received his Omega in the post. The Omega was shipping.

MicroDigital's apologists were in full effect around the time, stressing that the machine is a 'soft-puter', in that it can be re-programmed easily to add new features and correct problems. Except this design element doesn't appear to have been exploited much by MicroDigital. MicroDigital have also over the years blamed their problems in getting ASICs produced for them and then later, 'incompetent' FPGA manufacturers.

By now, the Omega's StrongARM processor is officially out of production and their decision to use a custom, in-house designed graphics engine over an off the shelf chipset, appears to have been a grave mistake. While there is some concern over whether or not MicroDigital have enough resources to support their Omega userbase, the damage sadly done to their reputation as a result of their wild cat PR tatics cannot be overlooked. Take your pick from bizaare posts to usenet, the rollercoaster ride that was the MicroDigital news pages (since purged by themselves), or telling the press - the very people who will help promote your stuff for free provided it's all true - to get lost.

Despite the fuss and unrest surrounding the Omega, Desk, however, are pleased that their users are, reportedly, happy with their Omegas. Some users are running RISC OS 4.37 on their computers, for example. As Desk put it: "Reactions from users of the Omegas currently in the field are that the computer works fine."

Links
Desk website
MicroDigital website
Thanks to Stan Williams for his help in translating Desk's newsletter.

This article has been linked to, or is available in the following formats:

[Printable] [Digg this] [Blog search]

imj (-1.5) 18/2/04 3:30PM	"The computer works fine" .... "as a doorstop"
Pete 18/2/04 3:42PM	Do you have one?
maikl 18/2/04 5:05PM	I think that the article is way to negative. Hey, at least the computers get now delivered to end customers. The computer might not be totally ready yet but it can already replace the RiscPC. Now that the computers are delivered and so there are not many motherboards to be fixed anymore there might be much more time to concentrate on the network issue and the other anhancements. Why is this not mentioned in the article. And I think it is not necessary to mention all the past mistakes of MD? You don't mention the Castle/Linux issue in any Iyonix article, do you? We have now two new RISC OS computers. There may be reasons to choose the one or other - but this shoulnd't lead to a split in the userbase. So in my opinion a fair comment would have been better.
ksattic 18/2/04 5:28PM	*In reply to maikl:* I disagree that the Omega can now replace the RiscPC. Until it has working ethernet and USB, not many people are going to consider moving to one. The Castle/Linux issue has definitely been mentioned on Drobe before. Castle released the source to the code in question (the HAL) under the GPL. There's no need to put details of it in an IYONIX review as it is a non-issue when considering the purchase of one of these machines.
mrchocky 18/2/04 5:31PM	*In reply to maik:* much as I'd like to see otherwise, the Omega simply cannot replace my RiscPC, because it still lacks the crucial networking support that I need to do my work. You're right, it doesn't mention anything about Castle's brush with GPL - that's because the article is about MicroDigital, and not Castle, and because the article is a history of the Omega development, as well as outlining many issues that are still current (contrary to the above issue, which was resolved to the satisfacton of most people a year ago). Chris tried hard to be positive in this article, but as we can clearly see from MD's coloured history, this is rather difficult, in light of the continuing status of this most controversial machine.
diomus 18/2/04 5:35PM	I don't believe the article is too negative: we've offered different views on the Omega and tried our best to be fair. For some people, the Omega cannot replace the RiscPC just yet, as it doesn't have USB or Ethernet support (for example). Once these are cleared up, we expect it will do. How much time is being spent on correcting motherboards hasn't been disclosed so we don't know what development time is being right now on the Omega. Also, there hasn't been a good round up of the Omega's history for some time so this seemed a good place to mention it. Actually, whenever the Iyonix or Castle come into contact with Linux in some way, we usually mention their past engagement with it. Also, (from what we've gathered) Castle have since re-wrtitten the code that was 'based' on the lifted GPLed kernel source, so it's perhaps not an issue anymore. I'm sorry that you don't feel this article contains 'fair comment', because like with all our articles, we try our very best to be as fair as possible. Chris.
GrahameP (-1.5) 18/2/04 5:54PM	When will Microdigital be in a position to return my £250 deposit that I asked them for in October last year and chased at least twice with no reply?
GrahameP 18/2/04 5:57PM	P.S. I also want the credit card interest on the deposit over three years (minus two months). I've been made redundant and I need the money!
robert79 18/2/04 6:18PM	Why don't you just get the credit card company to sort it out?
GrahameP 18/2/04 6:20PM	Because it has been too long for them to be interested - looks like a solicitor's letter as the next step.
imj 18/2/04 6:26PM	You should get a better credit card.
blahsnr 18/2/04 6:44PM	Or buy products that actually exist in terms of the published specifications.
dgs 18/2/04 6:44PM	*In reply to GrahameP:* As imj implies, it should not be "too long" for the credit card company to be interested - you've apparently paid MicroDigital 250ukp and got nothing in return. Did the credit card company themselves actually tell you that? From my experience, MicroDigital have normally returned people's deposits if telephoned regularly, or when presented with a deadline after which you will resort to a small claims court action. (Neither of these options need you to pay a solicitor to write a letter). Registered letters requiring proof of delivery have apparently been turned away by MicroDigital in the past, so that route may just waste you time and money. If you placed your order directly with MicroDigital, rather than through a dealer, you're fortunate in a way, because you don't have to put up with them "passing the buck" as to who is responsible - as I did. dgs
dgs 18/2/04 6:59PM	*In reply to maikl:* "Now that the computers are delivered and so there are not many motherboards to be fixed anymore" Are you suggesting that the vast bulk of Omegas sold will be from people who put down deposits after seeing the (wildly inaccurate) brochures from two or more years ago? Don't you think that, when the RISC OS public sees all these many Omegas being demonstrated on "the user group circuit", they'll be so impressed that there will be a flood of new orders? (Let's look at the positive side of some of the criticisms. For example, no working network card = not usually connected to always-on broadband connection = less security risks ) Incidentally, I see more of the "user group circuit" than just about anyone else, and I've still never managed to see an end user bring an Omega to a user group meeting, despite considerable efforts to encourage people. (Presumably Desk were referring only to the user group circuit in mainland Europe). At ROUGOL we've never seen an Omega at all (end user or otherwise) - the one Omega that was due to appear died before it even got there. dgs
maikl 18/2/04 7:29PM	*In reply to diomus:* Yes, you are right on some point you say. But somehow, when I read the article, I found it very negative. For example the headline "Omega production saga continous". I don't agree here - the saga does not continue as normal as there is a real difference now: End users get the product they are waiting for since 3 years. If you just read the headline you could think that the computers are still not delivered (or delivered shortly). But in general I always enjoy to read drobe.co.uk so keep up the good work.
maikl 18/2/04 7:35PM	*In reply to dgs:* Yes, I think many people ordered their Omega already. I did because there was a hugh increase of the price so I wanted to get a "cheap" Omega. And the people who did not order now may want to wait until Network/USB/Graphic acceleration is working. But yes, I saw the Omega on a user group meeting in Germany. I havn't seen a Iyonix yet (that's mainly because I am now in the USA for a few month) but I am looking forward to that as well.
blahsnr 18/2/04 7:59PM	In reply to maikl If I had waited until now for an Omega I would be hopping mad that a large proportion of the features listed on the MD page entitled 'Omega Computer Basic Specifications' appear to still not be implemented. It is true that something called 'Omega' has been delivered but it doesn't seem to be the same as the computer described on the MD website. As for Network USB and Graphics acceleration these features are already available for three RISC OS computers and have been available for at the very least 14 months. The computers are the RiscPC/A7000+ (via ViewFinder NIC's and USB podules) and the Iyonix.
TonyStill 18/2/04 10:45PM	I'm puzzled on the Ethernet issue. The 100Mb/s E/net NIC in my RiscPC works fine at 100Mb/s. How come RO4.03 on a RPC can cope if it's an OS issue? The RPC can't cope with sustained traffic at that rate because, as I understand it, it overwhelms the bus. However, my other RO computer can cope with sustained traffic at that rate, so it's not really the OS, is it?
zakalwe 18/2/04 11:18PM	If you're bad at designing computers, do not try to build and market one. If you do release a computer, make sure it works as advertised, because you probably will not get a second chance. Of course there always will be things that go wrong, but deal with them swiftly and openly (even if it costs you money) in order to save your greatest asset; your clients. (sorry, just my personal opinion)
dgs 18/2/04 11:29PM	*In reply to zakalwe:* But MicroDigital did get a second chance after the Mico! I've sat and listened at user group meetings (partly organised by me, promoted by me), as people who bought the Mico asked when the features they paid for, would work! Do I need to give prizes for people to guess what David Atkins said in response to those questions? It probably wasn't funny then. It's far from funny now. Is anyone still fooled by it? dgs
dgs 18/2/04 11:31PM	*In reply to maikl:* Could I just confirm, have you received an Omega? dgs
imj 19/2/04 12:27AM	*In reply to TonyStill:* That's the point... MD are spreading FUD about the OS which is blatently untrue. Blaming RISC OS for them being unable to produce their working ethernet is just nonsense. The TRUTH is that the RISC OS network stack, as stands, only has a throughput of a few MB/sec - with only 10baseT cards at best available until quite recently, that was sufficient. This could be improved by updates to the Internet module to run at higher rates, but the point is the achieved DATA rate is of no consequence to the ethernet SIGNALLING rate (ie 10baseT/100baseT/Gigabit) as MD would like to have everyone believe. Has anyone asked them the simple "If Castle and Simtec can make a 100baseT card, why can't you?". Hell, even a 10baseT card would be better than nothing on Omega, now wouldn't it? The saga rolls on and as I see it, the more Atkins tries to bulls*** his way out of this hole, the deeper he ends up digging. If it wasn't such a shame for this market to waste its time on this nonsense, it'd actually be exceedingly funny to watch.
maikl 19/2/04 1:23AM	*In reply to dgs:* No, I haven't received mine yet but I got an e-mail from my dealer to confirm delivery. But I am a special case because I am in the USA and the computer gets send to Germany so any delivery between today and end of August is the same for me. And I really hope that at least network is working then USB I don't really need now.
guestx (-1.0) 19/2/04 10:50AM	*In reply to maikl:* "And I think it is not necessary to mention all the past mistakes of MD? You don't mention the Castle/Linux issue in any Iyonix article, do you?" *In reply to mrchocky:* "You're right, it doesn't mention anything about Castle's brush with GPL - that's because the article is about MicroDigital, and not Castle..." *In reply to Advice:* read thoroughly first, then post! But aside from such slap-downs, doesn't this underscore the need for small producers to use as much generic hardware as possible, even though they don't get to show off sexy new designs? Of course, the most extreme approach is just to use the most generic of hardware - the Intel-based PC - and put emulation on top.
blahsnr 19/2/04 10:59AM	Re Ethernet. Net100 cards do show speed advantages over older 10baseT cards in RiscPC's. Even if MD's alleged comment is technically correct it would not explain the complete absence of a released driver merely why it wouldn't perform better with a faster card. Or am I missing something?
nunfetishist 19/2/04 11:29AM	Even from a simple point of view, I would have thought even if RISC OS couldn't receive data at 100Mbps because of the way it collects data from it, it might be able to send the data at close to that, or at least send it at a lower latency. This of course assumes that the networking stack sends a packet as soon as its got it, rather than waiting for the centisecond timer again. So they're worthwhile having over the older (and more expensive) 10Mbps ones.
Stewy 19/2/04 2:06PM	"There's also a sticker showing copy '300060' of RISC OS 4. The cynical among us will presume the 3 refers to the serial number subset allocated to MicroDigital and that 60 is the unit number for that ROM. And, I suspect, the 'cynical' have guessed wrong... My serial number for Virtual RiscPC also begins 300xxx. <ridiculous rumour alert>Hey, maybe the Omega is using Virtual Acorn software also?!</ridiculous rumour alert>. Nope, just checked my shiny Omega, and that's not the case "delays in producing the Omega were down to testing and fixing completed motherboards." Delays are never great. Surely it's a very good thing though that MicroDigital are sending out machines that have been properly tested, and avoiding the fiasco of having a machine that needs to be 'neutered'/'fixed', or whatever?! Anyway, it's encouraging to have a choice of new machines to run RISC OS natively.
blahsnr 19/2/04 3:23PM	stewy Problems leading to delays are an inevitable part of any project. As they are inevitable you plan for the unforseen and bear in mind that Murphy and his law can strike at anytime. However if everything you try and do is fraught with problems you have to ultimately ask yourself if there is not something wrong with your approach, and/or if the project is simply too ambitious for the resources you have at your disposal. Even when a project can be considered finished (ie machine running according to advertised spec for example) you can always improve matters. Do you believe that the RiscPC was 'insufficiently tested' because some users reported much greater stability on their machines when the C32 fix had been carried out?
epistaxsis@work 19/2/04 6:11PM	what is more worrying is that the photos show the motherboards on top of static bags which is not good how long will your omegas last..? unless of course MD have perfected some new static removed 'twister technology
stdevel 19/2/04 7:24PM	Pink bubble bags are conductive throughout. Surface resistivity on all sides is usually around 10^10-10^12 Ohms/sq. One might even propose that the outside is "safer" because of lack of bubbles and ergo a larger number of contact points. Removing a device from a bag and placing it on top of that same bag and then onto a surface is about the safest operation you can perform. Unless.... you drop the bag Let not that, nor worry as to whether Frank is competant enough to work a wrist strap stand in the way of a healthy considered debate though......
Use the forum for more comments on this article