StrongARM card turns ten years old | RISC OS news and information from Drobe


Beta! \| About us \| Contact \| Submit news \| RSS \| Twitter	Webspace \| Tech docs \| Downloads \| BBC Micro \| Gallery \| Wallpaper

StrongARM card turns ten years old

By Chris Williams. Published: 26th Mar 2006, 23:30:34.

And a decade on, we're still using 'em [Updated]

On this day ten years ago, the first prototype StrongARM processor card was powered up by Acorn engineers. The experimental kit managed to run at a cool 228MHz, running software nearly six times faster than the 40MHz ARM710 processors used in RiscPCs at the time. The card, which drew one watt of power, would later go on sale in September 1996.

"Everyone was amazed with the speed when it first came out," said Chris Evans of dealer CJE Micros. "By the time the first delivery arrived we had back orders of well over 100. They have sold steadily ever since."

Steven Pampling was one of the first to receive the card from Acorn, who offered it initially to Clan members, and immediately upgraded his RiscPC 600 machine.

He said: "I compared it to driving a car with the handbrake on and then suddenly someone showed you how to remove the handbrake. You were used to clicking and having a slight pause, suddenly it just happened."

Using Impression and ArtWorks, real ale fan Steve would produce a guide to local beers on his RISC OS computer.

He said: "At a time when people told me that you needed a full professional printer's DTP setup costing tens of thousands of pounds, I was exporting from a Squirrel database straight into Impression and putting together multi-page items with multiple columns and entries not splitting over column or page boundaries. With the StrongARM that took a few minutes to feed the data to the PostScript file - I still haven't seen a PC based setup that can do the same."

Steve recalled how various applications, such packages developed by Computer Concepts, had to be patched in order to work with the StrongARM card. He now uses an Iyonix, alongside his StrongARM RiscPC, A5000, and a PC laptop from work.

Steve said: "All in all I expect that in five years time either RISC OS will use a HAL on a motherboard with a zippy fast processor, or we will all be looking back on an interesting OS that isn't available anymore."

VirtualRiscPC and Iyonix user Dr Peter Young held off upgrading until 1997. He said: "As I'm not technically knowledgeable, my main memory of upgrading to StrongARM, and before that, from RISC OS 2 to 3 on an A420/1, was of the terror inspired by having to delve into the computer and the extreme relief that it all worked OK afterwards.

"I was impressed by the speed increase, but not, as I remember, as much as I had expected to be. Certainly not as much as I was by the increased speed of the Iyonix."

Tony Still, author of OHP, also picked up a card from the Acorn Clan, and remembered a website being set up around the time to track the distribution of the upgrades to end users.

He said: "We also had the fun of StrongARM-incompatibility with software. A bit like the transition to 32-bit that we've had with the Iyonix and the A9. The StrongARM CPU architecture has a different cache arrangement that broke some software - seemingly there is always a price to be paid for progress.

"I think the StrongARM was essential in giving the RiscPC its longevity. It took the performance right back into contention when Acorn seemed to be being left behind and gave us a machine with quite a respectable absolute performance. I can still bear to use my RiscPC whilst I couldn't cope with anything slower; I don't understand how people cope with non-StrongARM or pre-RiscPC computers."

A StrongARM card and RISC OS 3.7 ROM set would set you back almost 300 quid. The public got a glimpse of the pre-production card at the WROCC's Wakefield show, held in May that year.

WROCC event organiser Chris Hughes said: "The StrongARM card was shown to the public at the first Wakefield Show, at the Cedar Court Hotel, where it was unveiled. It was in a glass display, but was still the prototype."

Third party companies, such as Castle and APDL, later produced their own StrongARM card upgrades, by driving the processor at a much faster speed or strapping extra memory to the daughter card.

It was at 11.12am on March 26 1996 that boffins working for Acorn's technology division, ART, switched on the pre-production StrongARM chips from Digital Semiconductor. In February 1995, Digital licensed the blueprints of the ARM processor designs to produce a family of high performance 32 bit RISC chips. The move came about the time electronics industry analysts began realising that the ARM architecture was ready to take off in a big way. A year later in 1996, Digital presented the SA-110 StrongARM to makers of PDAs and set top boxes with predictions that they will sell billions of the new processor by the end of the century.

Digital later handed over the StrongARM designs to Intel as part of a lawsuit settlement; the architecture was then grown into the XScale family by Intel, who at the time fancied the idea of using RISC ARM-compatible processors for particular applications.

A poll in August 2005 revealed that 65% of drobe.co.uk readers still actively used a RiscPC - a computer which itself turned ten years old in 2004.

For a further nostalgic trip, see these two official Acorn documents about the StrongARM and the changes it brought about.

Links
StrongARM developments from Robert McMordie's Acorn history guide
RiscPC history
StrongARM processor photo by Richard Olivey

Discussion

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

Happy birthday StrongARM.

I can remember the time I upgraded from a 600 chip to the StrongARM chip, I kept on double clicking on apps thinking that they were not loading, then I looked on the iconbar and found about 10 copies of the same app, I never saw them load in it was just so much faster.

Revin Kevin on 27/3/06 12:41AM [ Reply | Permalink | Report ]

I waited till 1998 to get my first RiscPC, an ARM610 model from CJE. The main difference for me when I got the StrongARM card the following year (rev J, 200MHz) was the ability to play mp3s! Aside from that, the machine just came to life. The difference in speed was like night and day. Then in 2000 I invested in a Kinetic card and the speed difference again was outstanding.

Upgrading again to an Iyonix in 2002 gave another massive speed increase, of the kind I saw when upgrading from the ARM610 to the StrongARM.

ksattic on 27/3/06 4:58AM [ Reply | Permalink | Report ]

I can remember receiving the order card ten years ago, and then the package arriving on my doorstep. The speed increase was uncanny. I also remember my nervousness at my attempt (which was successful) at cutting the tracks behind the dip switch, allowing an overclock to 287 MHz

terrahawk on 27/3/06 11:35AM [ Reply | Permalink | Report ]

When the StrongArm (plus RISC OS 3.7) was first offered by Acorn, it cost just 100ukp, not the 300 mentioned above.

nikgare on 27/3/06 12:02PM [ Reply | Permalink | Report ]

nik: ISTR that there was a special offer to schools of a SA upgrade for 99GBP. I thought the offer cut about 100 GBP off the price (not 200GBP). Were you teaching at the time?

My school agreed to two - providing I paid for one of them! I took the one with me when I left. ;-)

jc on 27/3/06 12:11PM [ Reply | Permalink | Report ]

Yes, I recall paying 99 quid for mine, which I seem to remember required me to join up to the Clan.

That was an exciting time - first of all upgrading from an A410 to a RPC 600, where most of all the colourful graphics impressed, then a month or two later to a SA110 (which I'd ordered at the same time as the RPC) and getting another speed boost which was almost as big again.

Shame the Pheobe never drove the chip properly. I still maintain I'd be perfectly happy using a system based on the chip for day to day work if disc and memory accesses were speeded up. It's stunning that they managed to get such a fast chip running in a system that was designed to cope with clock speeds up to only half as fast, if that.

ninja on 27/3/06 12:41PM [ Reply | Permalink | Report ]

"I still maintain I'd be perfectly happy using a system based on the chip for day to day work if disc and memory accesses were speeded up" isn't that what the omega is for?

if you can get one, and get it to work of course.....

still makes me chuckle when i see a cellphone with a strongarm, or a raid card with 4 xscales.

simo on 27/3/06 1:24PM [ Reply | Permalink | Report ]

I took advantage of the release of the StrongARM card to move from an A3010 to a RiscPC. However, I had to wait a few weeks after receiving the RiscPC for the upgrade, due to the demand. So I went from one quantum leap in speed to another in very short order. I had eight very reliable years of use from the StrongARM card, upgrading to ROS4. No my mother uses it, and is very happy with it.

JWCR on 27/3/06 1:43PM [ Reply | Permalink | Report ]

I think a lot of us have stuck with the StrongARM because none of the new machines seem to do a lot new. Just the same things a bit faster. Of course, if you haven't upgraded to SA yet, and you still use RISC OS, you really should join the club. The speed and functionality increases are rather good indeed. -- Spriteman

Spriteman on 27/3/06 4:02PM [ Reply | Permalink | Report ]

Spriteman: New machines such as the Iyoinix do have a lot new; you have PCI cards, USB devices, and graphics of up to 2048x1536x32bpp at high refresh rates with quick redrawing. Plus the "bit faster" isn't so much the 3x or so of the XScale over the StrongARM, but the 5x speed of the memory, and 20x speed of the disc interface, which make a real difference to the feel and usability compared to previous generations.

druck on 27/3/06 4:33PM [ Reply | Permalink | Report ]

And TV on the desktop, very fast mail processing...

DaveW on 27/3/06 7:18PM [ Reply | Permalink | Report ]

Without the migration to Iyonix or the latest pure RISC OS machines is what is likely to hasten the end of our system.

Imagine what could be done if a few thousand of the diehard A3***, A4***, A5***, RiscPC users migrated to current hardware. Support by use is fine - BUT the survival of the system requires the cold hard cash maintaining current hardware and software provides.

The rate of new hardware/software on our platform is not so frequent that people can't make a move every two or three updates! Divide the cost by the number of years and so many users are only paying chip money. Updating would bring it to beer money (not quality champagne)

We have a champagne system which so many are only prepared to pay chip money for. Come on update (for beer money) and we might get to keep the champagne flowing. Don;t wait until the bottle is empty.

rmac on 28/3/06 1:46AM [ Reply | Permalink | Report ]

Rmac: I totally agree. Probably the biggest fear I have at present is that there may never be a replacement for the Iyonix, with the current state of the market. The amazing longevity of the RiscPC is a testiment to its future-proofed design, but unfortuately it is now hurting current hardware manufacturers.

timephoenix on 28/3/06 8:35AM [ Reply | Permalink | Report ]

Most people aren't going to buy a new machine when they can emulate one faster, for less money (similar thread going on elsewhere). It's not the longevity of the old machines that's holding us back now - they really are very long in the tooth now, but the lack of anything people think is really worth upgrading to.

SimonC on 28/3/06 12:48PM [ Reply | Permalink | Report ]

The RISC PC wasn't that future-proofed. It's old and slow now when compared to modern computers. However, we don't really have software that makes enough of a demand on the RISC PC that people would want to upgrade. If there was a games market that would surely have produced the need for something faster. Likewise with video processing, etc. In short, if the software is no more demanding than 10 years ago the computer is fine for the job it does. -- Sprite

Spriteman on 28/3/06 1:11PM [ Reply | Permalink | Report ]

Now that would be nice. A new game that makes full use of the extra speed and graphics available in both the A9 and Iyonix.

sa110 on 28/3/06 2:34PM [ Reply | Permalink | Report ]

Spriteman: sorry, can't agree there. If all you do is the occasional browse using O1 or Fresco, or the occasional letter using Impression or OPro, fair enough. If on the other hand you have a digital camera, or wish to use a modern browser like Firefox, or a display larger and deeper than 1024 x 768 x 16-bits, or work with any graphic files larger than about 15MB (sprites) or 5MB (jpegs or tiffs), or process a PDF file, or print out the results of any of the above quickly (especially in colour) or use any USB device, in short, do any of the many things a modern system is capable of, you'll find the RPC, compared to the Iyonix, is either dog-slow or incapable. The benefits of the Iyonix, as pointed out above, are more than just the faster chip: the entire system is more capable.

bucksboy on 28/3/06 5:59PM [ Reply | Permalink | Report ]

True, but have you recently used a modern computer with MacOS X, WindowsXP Pro x64 or Kubuntu? Neither the A9home (of wich we have one, too) nor the Iyonix can remotely compete with such systems. Modern RISC OS computers might have a slightly more productive GUI, but the other systems are not that far behind anymore and are continuing to catch up. And they offer a lot more software, wich is usually cheaper, better and faster (because of the faster hardware, but still) than the RISC OS equivalents.

Plus they run on much faster and cheaper hardware. For the price of an Iyonix Panther, you can get a PC with twice the HDD size, four times as much RAM (wich is also two to three times faster) and a CPU wich is roughly ten times as fast for integer and several thousand times as fast for floating point operations. The PC will also have a much faster graphics card with much more VRAM. Generally the software will also make much better use of the hardware's capabilities.

Of course if you are used to a specific software and have learned to use it efficiently, you'll want to stick with it, but that does not mean that it is better. You don't even have to look at different platforms for that effect. A StrongEd user will always say that StrongEd is better than Zap, until he really makes an effort to use Zap intensively for a few months. If he then decides that StrongEd is still better for him, fine. But RISC OS users usually don't make a real effort to get used to the work-flow of other systems. Many RISC OS users who think other OS are not as good as RISC OS have never really tried to get used to their way of working. But at some point, one has to make a choice just how much cash you want to spend extra on RISC OS so you don't have to learn to use different software packages on other platforms.

I really hate to write this stuff, because I really love RISC OS. I now work in the media industry, wich is why I had no choice but to use other systems, but over time I learned to regret that I had not switched earlier. For RISC OS to become an attractive platform for me again, it would have to leave the ARM architecture behind, become compatible with mainstream hardware platforms and massively widen the software base, by utilising open source software and porting WINE to RISC OS. Maybe an ARM processor could be kept on a PCIe card to speed up emulation for legacy software. IMHO the best path to achieve that would be to extend Virtual-RiscPC for Linux, so that more and more parts of RISC OS interface directly with the Linux kernel and run directly on the real hardware.

JGZimmerle on 28/3/06 11:54PM [ Reply | Permalink | Report ]

Oh yeah, I forgot: Feel free to moderate me down. :-)

JGZimmerle on 28/3/06 11:56PM [ Reply | Permalink | Report ]

JGZimmerle: my earlier post was intended to answer the assertion that a RiscPC was adequate for current RO computing. But I don't disagrree with your hardware comment: a large part of the current interest in emulation is presumably driven by the realisation that ARM chips have been outstripped by x86 development for desktop use. That said, I believe a lot could be done to improve the performance of current RO ARM hardware, by implementing disk cacheing and utilising the hardware DMA potential of the X-Scale, at least.

bucksboy on 29/3/06 8:19AM [ Reply | Permalink | Report ]

JGZ: I don't think it's fair any more to say that most RISC OS users are blind to other systems. Sadly, it's very hard these days to be in a position where you don't have to use them.

bucksboy: That might move it a little closer, but it'll still be a long way behind. Can't see disc cacheing making a great deal of difference on a system that doesn't thrash the disc all that often.

SimonC on 29/3/06 10:04AM [ Reply | Permalink | Report ]

Problem with disc caching is that when the system crashes, you could be left with needing to utilise discknight to sort your hdd out.

sa110 on 29/3/06 10:22AM [ Reply | Permalink | Report ]

sa110: True. If RISC OS was as stable as Windows or MacOS or Linux then it wouldn't be as much of an issue but currently a naughty application can take the system down quite happily. -- Sprite

Spriteman on 30/3/06 12:58PM [ Reply | Permalink | Report ]

disc cacheing: OK, but my main point (of which this was part) is that even current Iyonix hardware has got further development potential. It is remarkable how poorly the emulators perform in Dhrystone or MIPS terms compared with native hardware, and yet are as fast or faster than the latter for many tasks. This suggests (to me at least, but I'm just a user, not an expert) that a higher clock rate ARM is not the only way to achieve a worthwhile improvement in the performance of native hardware.

bucksboy on 30/3/06 8:56PM [ Reply | Permalink | Report ]

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

Create a new account
Forgot your password?

Search this website

This week's poll

Recent developments have left me feeling...

Assured ROS will appear on new hardware		55%
Pleased OS desktop features are being developed		10%
ROL and ROOL will eventually kiss and make up		5%
App developments are critical		10%
Dave Holden sleeps easy at night		19%

Discuss this | Archives