Intel trains XScale at future PDAs, mobiles

By Chris Williams. Published: 18th Sep 2003, 17:18:28 | Permalink | Printable

Don't over do it lads

A number of you have dropped us emails about Intel and their latest XScale plan so it's probably worth a mention - if not purely for interest's sake as it's a good idea to follow the progress of the processor architecture we're currently relying on and the corporation who's calling all the shots.

Over the past few of days, Intel held their Autumn 2003 Intel Developer Forum in California, USA and generally grabbed a lot of headlines. One mildly relevant piece of news to emerge is Intel's intention to take the XScale core and add their MMX technology to make a processor suitable for next generation PDAs and mobile phones. For next generation mobiles, think video capture support and similar gadgetry because merely phoning someone these days isn't enough - audio is so 1990s if not earlier. And yes, it's the same MMX that we saw introduced with the Pentium series, Intel's specialised instruction set for multimedia, 2D/3D graphics and now wireless support.

Intel as always are keen to focus on the low power usage of the XScale, something we've taken for granted for a long, long time and more importantly, it's interesting to see that Intel have found a future for the XScale which is thoroughly reassuring. Although other ARM cores exist, the XScale IOP range has shown to be the most suitable for RISC OS and if we're relying on an architecture over which we have no control or influence (unlike the mid Acorn era when ARM and Acorn were drinking buddies), it's therefore a good idea to pick a stable platform with a future. It's either than, or go for the emulation option on an architecture that is guaranteed a future.

The aforementioned newly announced XScale with MMX processor, code named Bulverde, isn't going to be suitable for desktop use, of course. However when Castle bought RISC OS from Pace, they were quick to state that they wanted to push RISC OS into non-desktop devices.

Once upon a time, it was the in thing to deride Chipzilla and their storage heater worthy Pentium chips while smugly displaying 'Intel outside' stickers against garish backdrops on humble homepages. How times change.

Links

Intel Intel shows off next-gen XScale phone chip from The Inquirer

Previous: Select 3i3 quickly follows 3i2
Next: RISC OS P2P client now with upload support

Discussion

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

Oh it's good to see they've found something else to keep themselves busy with, rather than just making Pentium chips!!!!!!!!!!!

Timothy609 on 18/9/03 6:11PM
[ Reply | Permalink | Report ]

Yeah, network cards, switches, chipsets, wireless infrastructure, motherboards, BIOSes, compilers, debuggers and other development products, etc....

nunfetishist on 18/9/03 8:11PM
[ Reply | Permalink | Report ]

Hey I made that "Intel Outside" logo as part of one of the ArtWorks competitions!

-- C'mon, mod me down, PUNK!

simo on 18/9/03 9:22PM
[ Reply | Permalink | Report ]

*used to have an intel inside sticker on his radiator*

AndrewDuffell on 18/9/03 9:57PM
[ Reply | Permalink | Report ]

*Whatever* happens, people should never write RISC OS software that takes advantage of these Intel extensions. Because then you'll be locking yourself into Intel, which for years RISC OS advocates having be trying to find ways to avoid. Just stick to plain boring ARM code, then if things go pair-shaped in the future, then you can go and use a real ARM.

nunfetishist on 18/9/03 11:34PM
[ Reply | Permalink | Report ]

I disagree. Ignoring half of the instruction set is not progress. Why do you think I sold my kit and bought a cheap (really cheap) PC? Has a lot to do with the fact that I can surf the net, watch a film clip, listen to music and still be able to encode an album to MP3 at about one fifth of real time. How does it achieve this? Well I'm afraid it's those disgusting evil multimedia instruction set extensions...

Why should anyone want to run RISC OS on their PDA, phone, desktop computer or whatever if it and the apps deliberatly hobble themselves by not using the full power of the hardware?

I say go use the extensions - they're bloody useful!

robert79 on 19/9/03 12:51AM
[ Reply | Permalink | Report ]

Provided you could have a software MMX emulator like the FPEmulator, I see no problem in using MMX instructions.

The only thing is, I really can't see how MMX instructions (which presumably are variable length) can fit into a RISC instruction set, consisting of fixed-length instructions. I should probably RTFM. -- Simon Wilson, Boulder, Colorado

ksattic on 19/9/03 3:05AM
[ Reply | Permalink | Report ]

Couldn't software be compiled for different processors, like on the Amiga. Then one version would be optimised for MMX processors but another version that still runs on basic ARM code. It would also work as an insentive to upgrade in order to improve the performance for that software by enabling the MMX extentions. Also, are there any other new instructions that ARM have added since 610/710/StrongARM? I find it hard to believe they just left it! -- Smiler -

Alex Melhuish

Smiler on 19/9/03 7:09AM
[ Reply | Permalink | Report ]

Smiler> Yes indeed.

But I think nunfetishists point was that if you become *dependant* on non-standard extensions then if Intel change or drop these (or simply lose interest in the ARM) you could be left in a bind.

As to ARM610/710 yes *ARM* have added extensions since (such as the VFP instructions on ARM10, CLZ and other odds and ends bits and pieces). Point is ARM define the standard and *all* ARM licensees can use the extensions. With MMX that's an Intel creation and whether it *could* be used by other ARM licensees might be an issue.

-- Annraoi

AMS on 19/9/03 11:10AM
[ Reply | Permalink | Report ]

Of course, if Intel have certian patents or copyrights on XScale MMX's instructions, it might cost a fortune to licence them to make an FPEmulator or similar to run them on machines that don't have them.

nunfetishist on 19/9/03 12:26PM
[ Reply | Permalink | Report ]

Coming back to the Intel Inside/Outside comment; Haven't Acorn computers had Intel chips of some description in them for years? I'm not talking the processor but rather one of the support chips on the motherboard. -- Spriteman.

Spriteman on 19/9/03 1:04PM
[ Reply | Permalink | Report ]

Does it *really* matter who makes the chips? Isn't it far more important that the pathetic "Acorn vs Microsoft/Intel" attitude is forgotten. That battle was lost at least 15 years ago. It would be perfectly feasible for RISC OS to exist as a fairly large niche market, but that aint gonna happen when such old-fashioned attitudes hang around. The market should be trying to embrace new technology, not shy away from it because it might cost some cash.

robert79 on 19/9/03 2:36PM
[ Reply | Permalink | Report ]

>Robert79

Microsoft and Intel have "monopolist" tendancies. One of their standard methods of attacking the competition is "embrace and extend", they appear to support an "open" or industry agreed standard - then add new features (which are *their* exclusive Intellectual property) and close out the competition.

ARM because it has many licensees can't be seen to discriminate against one over the other - so that ARM's standards generally are available to all and embraced by most. If Intel start adding Intel specific bits to the ARM standards that may "lock" companies into only using the Intel variant and gradually close off the competition.

All of this *does not* relate in any way to an "Acorn vs Microsoft/Intel" argument but rather to the broader notion that computer manufacturers should all compete fairly.

-- Annraoi

AMS on 19/9/03 3:24PM
[ Reply | Permalink | Report ]

In reply to AMS:

If the Intel extra features mean that the Intel ARM chips are better why shouldn't people use them? If other ARM chip producers want to compete they'd better provide something worth buying instead.

In reply to Robert79: The article itself noted how these old, anti-Intel views have been mostly lost in the past. "How times change." -- Spriteman.

Spriteman on 19/9/03 4:37PM
[ Reply | Permalink | Report ]

In reply to Spriteman (and AMS)

The thing mention by AMS must indeed be seen in broader scheme. Intel are indeed trying to DESTROY ARM Ltd. This should be obvious and has been told by me for ages since Digital sold off their cpu-rights to Intel. That moment I knew we'd be in BIG s***. And trust me. We will.

ARM cannot add new (useless) stuff to their core. It's a matter of their licensee in the case they want to use the Core for to add new items to it. So ARM will not take drastic measure in this matter. But if some chip manufactor wants to use that core in a video-encoding chip it might ad it owns multimedia-instructions to it and sell it as it's own ARM-based cpu. In this case ARM can't take actions against Intel for this. But it'll certainly undermines ARM strategy.

Also. I am STILL Anti-INtel AND anti M$. Just because of their monopolistic nature. The same goes for Apple and Adobe. It appears common for big corporations to bulls*** everyone around!

epdm3be on 19/9/03 4:55PM
[ Reply | Permalink | Report ]

It's very easy to be sneering about people who are anti-Microsoft/Intel. It's easy to paint it as being misguided and childish; just as it's easy to paint anyone who genuinely cares about something in that way; they leave their emotions wide open to be ridiculed. People who criticise such things, however, tend to believe they are above such things *purely to make themselves feel superior*, in my experience.

(An unhelpful comment, I know, but I can't help but get irritated by things like this. Goddamn it, it's not bad to dislike Intel and Microsoft.)

moss on 19/9/03 5:41PM
[ Reply | Permalink | Report ]

(I'm talking about people who try to put themselves above such matters, incidentally; not those who simply take the other side, or those who *genuinely* have no interest in such matters. But I've just seen too many forum arguments where people try to take the high ground purely to make themselves look grown up.)

moss on 19/9/03 5:44PM
[ Reply | Permalink | Report ]

In reply to epdm3be, ARM arent' exactly innocent when it comes to bullying people about with licencing and patent threats either. You make ARM sound like a nice little friendly couddly thing, which they're not. They're a company just like the rest. The difference is that Intel is a wee bit bigger

Like Spriteman says, who cares at the end of the day? All you want to do is run software to enable you to do stuff, do you really care who provides the processor on which it runs?

To satisfy moss and his "everyone has an opinion or they're just trying to show off" clause, I really think the ARM instruction set as it stands is a bit dated for desktop use. You certainly in this day and age want something with at least a few more registers in there

But I've had this argument here already

-- Dougal

Dougal on 19/9/03 6:53PM
[ Reply | Permalink | Report ]

Dougal: Using ARMs are desktop use was dated 10 years ago, though! Apple did what Acorn should have done at the same time: Move away, and provide an emulation layer for old software.

nunfetishist on 19/9/03 7:22PM
[ Reply | Permalink | Report ]

nunfetishist - well quite. I was just trying not to be too inciteful

-- Dougal

Dougal on 19/9/03 8:41PM
[ Reply | Permalink | Report ]

A number of points:

(1). x86 Have *far fewer* registers than the ARM, yet no complaints (curious).

(2). The ARM instruction set as far as it goes is fine (ok, there'd be one or two instructions I'd like added but I'd live without them). You could bloat out the ISA to the point where the chip would take longer to figure out "which" instruction was being asked for than the time taken to actually execute it and you'd wind up with a CISC chip like in the 1980's but where would be the point in that.

(3). RISC OS runs best on ARM, no emulated system yet equals the performance of ROS on an actual ARM and I don't think one will be likely in the next 3 to 5 years (by which time the ARM will have moved on and the gap will have been restored).

I'd agree with Dougal that ARM are not exactly all sweetness and light when it comes to potential patent infringments (I can think of BlackARM and PicoTurbo as examples...). They (like Intel) do tend to get a bit "heavy" if they perceive their Intellectual Property as being threatened.

The big difference is Intel controls possibly 90%+ of the processor market, ARM doesn't. Any excesses or pressure Intel applies is far more telling (and more damaging) to competition.

That having being said I am actually sorry ARM doesn't have competition as Intel does as it would force them to upgrade their processor performance in a way AMD forced Intel to do in the x86 field - and we'd all benefit.

-- Annraoi

AMS on 20/9/03 11:46AM
[ Reply | Permalink | Report ]

AMS: So Intel have 90% of the market and yet ARM have no competition...

Depends on the market, doesn't it? ARM have seemed very snug in their market of being an embedded systems (PDA, mobile phones, etc.) market, and they didn't do that be resting on the laurels.

As for the register argument, we've been there before here, and we didn't agree then, so I can't see how we're going to agree now, so I suggest we not rehash it. It'll only lead to an increase in my blood presure

-- Dougal

Dougal on 20/9/03 1:33PM
[ Reply | Permalink | Report ]

In reply to AMS:

I don't think you quite get what we mean when we mention emulation. This is not emulation as in Virtual Acorn where the whole system (processor and other hardware) is emulated. You take RISC OS 5 with its hardware abstraction. Write drivers in native code so that it can access the new hardware. Then you analyse the OS, find which routines are called the most and rewrite them in native code. The rest is caught by emulation of the old processor. As time goes on you continue to convert old code over to the new processor. The performance hit would hopefully be simmilar to that from the Aemulor on the Iyonix (allowing for the increased speed of a G5 processor over an XScale). Well, that's the plan in very general terms. The ins and outs go well beyond the scope of an article comment ;-)

Oh, and Intel doesn't control 90% of the processor market that ARM targets. :-)

-- Spriteman.

Spriteman on 20/9/03 1:42PM
[ Reply | Permalink | Report ]

Yes the 90% comment does seem to be a bit inaccurate.

I remember reading an article a while back, which stated that in 2001 there were more ARM processors sold in the world, than any other type of processor.

James

Walks on 20/9/03 1:51PM
[ Reply | Permalink | Report ]

MMX is more Archetecture modifications to allow media type alogrythums to run more efficiently, though what benifit they would have on the ARM I'm not to sure at all, personally I see the things introduced in the ARM 10 as being a lot more relevant, I just think intel wants to stamp MMX on the chip for marketing purposes, after all MMX... just as mobile things are going multy-media?

NoMercy on 20/9/03 4:28PM
[ Reply | Permalink | Report ]

*coughs* make that ARM 10 and 11

NoMercy on 20/9/03 4:31PM
[ Reply | Permalink | Report ]

Dougal>

Sorry Man, I wouldn't want to be responsible for raising your blood pressure. So I leave the register one where it is

Yes I take your point about the market. At one level Intel have the desktop market sown up, at another level ARM have the embedded processor market sown up. I think my point (perhaps unclear due to hamfisted writing on my part) is that Intel were forced to upgrade the Pentium because of competition from AMD. ARM has (so far) made sure that it has *no* compeditors that can offer an ARM instruction set compatible processor - and hence (perhaps) things have been left sit a little (IMHO). If ARM had had the same sort of competition as Intel had perhaps ARM would be aiming somewhat higher than the 400MHz they currently aspire to (as high end) - which some of their licensees (e.g., Intel and Samsung) have already shown is a breechable target.

Spriteman> The point is that *no* currently existing processor allows ARM code to be run as fast as the fastest ARM's currently available.

I don't quite follow your point about hardware abstraction (VA uses bog standard RISC OS which doesn't). Just because ROS 5 hardware abstracts does *not* necessarily mean you'll get better performance as the native OS it will call (hypothetically WindowsXP or Linux) would have *further* hardware abstraction for it's own purposes (you'd in effect be doing the hardware abstraction step twice - once in ROS (in emulated ARM code) and once in the underlying OS (in that machines native code)). Or have I missed something ?

-- Annraoi

AMS on 21/9/03 3:48PM
[ Reply | Permalink | Report ]

Arm do have some competition, but they seem to be well ahead, there push for low-power tech isn't to great for our desktops, but at least we can still buy our favorite CPU's somewhere.

NoMercy on 21/9/03 4:12PM
[ Reply | Permalink | Report ]

Hmmm. That was a rather embarassing and stupid rant I had up there. Sorry, everyone.

moss on 26/9/03 1:17AM
[ 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

The new apple of my eye
Would you swap your dusty Acorn for a polished Apple computer? Martin Hansen has been checking out the world of Steve Jobs and his range of shiny kit.

15 comments, latest by adh1003 on 6/1/09 1:06PM. Published: 17 Nov 2008

Random article

Windows nasties can affect ROS users too
It's true, it's true

25 comments, latest by quatermass on 12/3/04 12:11PM. Published: 5 Mar 2004

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
"On Drobe's view of confidentiality and 'public interest'... I find the words 'invasion of privacy' more appropriate"
Page generated in 0.084 seconds.