ARM — Heretic In the Church of Intel, Moore's Law

ericatcw writes "For 30+ years, the PC industry has been as obsessed with under-the-hood performance: MIPs, MHz, transistors per chip. Blame Moore's Law, which effectively laid down the Gospel of marketing PCs like sports cars. But with mobile PCs and green computing coming to the fore, enter ARM, which is challenging the Gospel according to Moore with chips that are low-powered in both senses of the word. Some of its most popular CPUs have 100,000 transistors, fewer than a 12 MHz Intel 286 CPU from 1982 (download PDF). But they also consume as little as a quarter of a watt, which is why netbook makers are embracing them. It's 'megahertz per milli-watt,' that counts, according to ARM exec Ian Drew, who predicts that 6-10 ARM-based netbooks running Linux and costing just around $200 should arrive this year starting in July."

Comments

Nonsense.

[Vectronic]

I don't mean to Dis-ARM, ARM or Armless...

But it will do exactly the same thing, 0.5 Watts now, 100K transistors now, 300 MHz now... it wont stay that way though, it's just a slimmer base to build upon, like using aluminum instead of steal. People will still keep reaching for the sky, and with a lighter structure, means they can reach even higher, even more MHz, more transistors, etc...

Re:Nonsense.

[AuMatar]

Not really. ARM has been around for a long time. Its biggest use is in embedded systems- phones, printers, etc. In those markets cost and power usage matter more than performance. They may make a line with more performance eventually, but they make money hands over fist in places where pennies matter (after all, if you sell 1M phones with a processor thats 5cents cheaper, thats 50K more profit). They won't give that up.

Re:Nonsense.

[Bill Currie]

Exactly, thus the MHz/mW phrase. Why use more electrical power than you need to? If you need more computing power, then build a bigger CPU using the same technology. It will still be more efficient, and that's the point: efficiency.

Re:Nonsense.

[Vectronic]

Yeah, but even your average phone is more powerful than your average PC was in 1982...

So "in the meantime" they will somewhat stick with the low+low, what happens when laptops, phones, GPS, etc all become the same device? People are content with the low power they have now, and with stuff like anti-jailbreaking etc, puts a limit on the push for better/faster/stronger because not many see it yet. People thought your body would fall apart at 50mph 100 years ago... "640kb ought to be enough for anybody"...

PSP (not the most relevant example) might be 300 MHz now, what about PSPII, still 300MHz? Doubtful.

Re:Nonsense.

[Vectronic]

I know, I was simply contesting the "goes against moore's law" part.

Coincidentally, that's when the greatest blasphemy to Moore's Law -- and the biggest threat to Intel's dominance -- is expected to make its entrance into the PC market.

When it isn't, it's similar to automotive racing, this seasons F1 has all sorts of new limitations on engine size, RPM, and materials to promote more power/speed out of smaller, doesn't mean they will stay constant at 750hp @ 12,000RPM, by the next couple of years they will likely be back up to 1000hp, just on a smaller platform.

In 10 years time, there will likely be some even more efficient processor out there (likely already exists) It's all effectively a part of Moore's law, the current base has certain limitations limiting it's linear climb, so a new twice as good base is developed to continue that climb.

Re:Nonsense.

[AuMatar]

PSP is a bad example. The biggest knock on it, and one of the reasons the DS won, was horrible battery life (although the main cuse of that was going with optical media rather than disks. Bad idea.

Phones might end up going up in power, but you miss the point. If they wanted megahertz, they could get it now. Better processors exist. The manufacurers don't want it- they prefer to save money and make a higher profit. There's billions of devices out there still using 8 bit microcontrollers. They'll never move to higher cpu power because its not needed- its a waste of their money (higher CPU power requires more transistors and thus more die space, for lower yield and higher cost). ARM occupies the niche above that- the devices that need more than the average microcontroller, but nothing as much as an embedded x86 chip. These are billions of devices per year, and they aren't going away. ARM may end up building higher CPU power chips as well, but they won't abandon the existing market.

Re:Nonsense.

[tlhIngan]

But it will do exactly the same thing, 0.5 Watts now, 100K transistors now, 300 MHz now... it wont stay that way though, it's just a slimmer base to build upon, like using aluminum instead of steal. People will still keep reaching for the sky, and with a lighter structure, means they can reach even higher, even more MHz, more transistors, etc...

You do realize that the Cortex series of ARM cores can get to around 1GHz, and that the Marvell (formerly Intel) XScale chips can scale to 1.25GHz easily. And that's when they're drawing a quarter to a half a watt. At worse, you're getting 1GHz/watt.

ARM is used everywhere, it scales handily from fleapower devices, to the GHz range used in the latest smartphones. For every x86 CPU sold, the PC containing it probably contains several ARM processors (Bluetooth and WiFi being extremely common peripherals with ARM processors). A cellphone usually has 2 - one driving the UI, and one in the radio, and maybe two more (again, Bluetooth and WiFi).

400-667MHz seems to be the "sweet spot" right now for a cellphone's ARM processor... (iPhone has it at 400-416MHz, the Palm Pre has a Cortex A8 at 667MHz). And the whole cellphone power management has to be able to drop power consumption to a mere 3 milliamps or so, including the power spikes to maintain a link to the cell towers.

Atom tries, but it's still an order of magnitude too much power for an entire system...

Re:Nonsense.

[Firethorn]

No one recorded the observations of standing outside in 50mph winds? Or of someone in free fall from a great enough height?

Go back a couple hundred years and people believed all sorts of weird things. Baths were bad. Bloodletting was good. The moon's made of cheese, earth's flat, earth's the center of everything, We can reach the moon/planets with a giant cannon, etc...

It was never really a widespread belief, if I remember right, the educated knew we'd be fine, more or less, and the truly uneducated didn't know what 50mph was. You had a selection of semi-educated people who would come to weird conclusions.

Heh, think of it as early scientific theories. They were made to be proven wrong(or not).

Re:Nonsense.

[AuMatar]

Not really. Brains don't spontaneously explode.

Re:Nonsense.

[BikeHelmet]

Looks like you got those numbers from the article rather than the spec sheet.

The power consumption listed is off a bit. That 100k transistor CPU only uses ~5 miliwatts load (less when idle), which is 0.005 watts - or averaged, ~0.002 watts.

However, most companies designing SoCs from it would embed tons of other stuff in the chip, like a GPU, USB controller, networking, etc. etc., so power consumption might increase to almost a watt when they're done, if everything is active.

I know Intel likes to boast about its roughly-one-watt CPUs, but they really have nothing on ARM as far as power consumption... and ARM has nothing on them as far as performance - but luckily Intel's stuff is so insanely fast that even at 1/20th the speed, speedy ARM SoCs are fast enough to run a desktop OS.

Re:Nonsense.

[NotInTheBox]

Go back a couple hundred years and people believed all sorts of weird things. Baths were bad. Bloodletting was good. The moon's made of cheese, earth's flat, earth's the center of everything, We can reach the moon/planets with a giant cannon, etc...

Even nowerdays people believe wierd nonsense and myths... like that one about how people once believed that the earth was flat.

http://en.wikipedia.org/wiki/Flat_Earth_mythology

What else that is commonly believed will turn out the be wrong?

Re:Nonsense.

[jimmydevice]

Religion, politics, benchmarks, the check is in the mail, I won't cum in your mouth.

Re:Nonsense.

[mattrumpus]

Go back a couple of hundred minutes and people believed everything was created by a big man in the sky!!

Re:Nonsense.

[itsdapead]

Yeah, but even your average phone is more powerful than your average PC was in 1982...

Actually, when the ARM appeared in 1987 it wasn't touted as a low-power chip, but was developed by Acorn as part of a desktop workstation chipset that could easily show the x86s of the day a clean pair of heels. One of the first products was actually an accelerator card for the PC.

(Sorry about the PDF links: the parent site is http://acorn.chriswhy.co.uk/)

The Archimedes/Risc-PC "workstations" stayed in production to the late 90s (and there have been Amiga-style holdout products until very recently) and were always decently fast - but they couldn't compete with the Wintel dupooly and started to lose out when FPUs and, later, accelerated graphics cards became the norm on PCs. By this time, ARM had been spun off and had (wisely) started to concentrate on embedded systems which (at the time) didn't need such things.

Re:Nonsense.

[Paul Fernhout]

Just for fun, theoretically, you can put payloads into orbit and on routes to the moon/planets/asteroids if you give them a solar sail. (People could not survive that trip, unless encoded in data bits and silicon.)

Maggots and leeches are proving effective in medicine in various ways.
    "Maggots and Leeches: Old Medicine is New"
    http://www.livescience.com/health/050419_maggots.html

In round figures, people are about 90% bacteria by numbers, and about 10% bacteria by weight. Bathing too often may disrupt your bacterial ecology and lead to infections or skin problems, and growing up in too clean environments may lead to immune problems. Although exactly what is too much is problematical. See:
    "The filthy, stinking truth: The messy history of cleanliness, and why our obsession with dirt may be making us sick."
    http://www.salon.com/mwt/feature/2007/11/30/dirt_on_clean/

Until we actually landed on the moon, the best scientists still thought landers might sink into dust. Someday, we may turn the Moon into a green paradise using greenhouses and artificial lighting or mirrors.

Psychologically, the individual's perception is still the center of everything (though people try to move beyond that in their thinking). Quantum mechanics reflects this. Still, we may be living in a simulation in which case, like those living in Plato's "Cave", most of what we assume may be just a shadow of the truth:
    http://www.simulation-argument.com/

Anyway, just having fun with your points. I like your insightful comment that knowing enough to be dangerous (as opposed to nothing or lots) is a source of difficulties.

Here is the big issue with Moore's law and it was forseen in the 1960s:
    http://www.educationanddemocracy.org/FSCfiles/C_CC2a_TripleRevolution.htm
"The fundamental problem posed by the cybernation revolution in the U.S. is that it invalidates the general mechanism so far employed to undergird people's rights as consumers. Up to this time economic resources have been distributed on the basis of contributions to production, with machines and men competing for employment on somewhat equal terms. In the developing cybernated system, potentially unlimited output can be achieved by systems of machines which will require little cooperation from human beings. As machines take over production from men, they absorb an increasing proportion of resources while the men who are displaced become dependent on minimal and unrelated government measures -- unemployment insurance, social security, welfare payments. These measures are less and less able to disguise a historic paradox: That a substantial proportion of the population is subsisting on minimal incomes, often below the poverty line, at a time when sufficient productive potential is available to supply the needs of everyone in the U.S. The existence of this paradox is denied or ignored by conventional economic analysis."

So, we are about to see a lot of divide-by-zero errors in economic equations as computing prices falling to zero drives almost every other price towards zero.

Re:Nonsense.

[fractoid]

Sorry to hijack, but your mention of F1 and their stupid technical limitations really hit a nerve. That's my biggest pet hate about F1 - they keep trying to slow the cars down and make them safer by imposing arbitrary limits on various components. What they should do is impose a cost cap on the cars. Do as much research as you want, but the car itself has to be buildable for $100k. The next year, drop that to $80k. Eventually you have a vehicle that is at once the pinnacle of automotive excellence, and available for a decent price from a dealer, and better yet - the technology would be easily adaptable by passenger cars. Antilock brakes, seat belts, traction control, most technology in modern cars was developed for F1. If we could speed up the transition process then that would be the best thing possible for the automotive industry as a whole.

Re:Nonsense.

[CronoCloud]

SP is a bad example. The biggest knock on it, and one of the reasons the DS won, was horrible battery life (although the main cuse of that was going with optical media rather than disks. Bad idea.

The battery life may not be as good as a DS, but it's still tons better than any of the other competitors Nintendo had in handhelds. The PSP is also much much more capable than the DS is, almost but not quite on the level of the PS2. Optical media was a good choice for it, because back in 2004 large capacity flash with as much capacity as a UMD has was expensive and you can stamp out optical media cheaply and in large numbers.

The DS hasn't exactly "won" the handheld war, it's the most popular, but the PSP is a strong second.

Re:Nonsense.

[Vectronic]

I agree and disagree.

F1 is top dog, so I don't think a monetary limitation should really take place. It's where most of the major experiments take place, so shouldn't really be limited on price because developing new materials is expensive, especially if it's rare/new, so they might be able to build most of the car for $60,000, but the new allow/polymers/etc that make up the frame alone might be another $60,000 by itself... instead of "well what can we do with $20,000? - hmm, carbon fiber and aluminum?" They need to build new machines that build the new machines, get/make the new materials which are basically one of a kind, etc, a set of tires is about $5,000 because they are so customized down to the molecular level.

F1 spends the money + time, which trickles down to GP2, A1... those trickle down into GT, Rally, StockCar, which have more "real world" limitations, those trickle down to 'supercars', and then down to normal passenger cars.

So I think limiting displacement, fuel type, etc is a better way to go so they can sort of invest/start it, but everyone else gets the benefits of the product (eventually). Theoretically your idea is what happens already, Mercedes, Honda, BMW, Toyota, Renault, etc all have road vehicles, so basically they are spending the huge amounts on F1 and GT, to create their road cars.

Side note, I'd like to see an official electric/alternative F1 start, instead of these little parade events once a year.

Re:Nonsense.

[mdwh2]

Nonsense. This is like saying a militant atheist is a "Christian", because someone splashed water on their head as a baby, they had christian parents, and were forced to go to "christian" school.

If we define one's religion not by that person's belief or identity, but by someone else's actions and definitions, then we get into all sorts of nonsensical situations. Consider, if I decide that me shaking your hand whilst I'm dressed up as a pirate makes you a member of the Church of the FSM, and that you're "not really permitted to leave", does that make it true?

There is no such thing as a muslim child, just as we would not talk about a Marxist or Keynesian child.

Although I note that whilst usually religious organisations promote these definitions in order to inflate their numbers and force religion onto children, I suspect you're instead taking advantage of anti-Islam viewpoints, in order to make Obama look bad. But I don't really care about the politics here, just your nonsensical non-consensual definition of labelling. Who cares if he is a muslim anyway? Christian or muslim, or Church of Intel, just so long as he keeps his religion out of politics, unlike a certain recent President...

Re:Nonsense.

[AuMatar]

100M sales to 35M sales, and the ratio is growing. DS won.

Too bad the CPU isn't the only thing drawing power

[anss123]

ARM chips are nice, but they are not as fast as Atoms and their low power usage does not guarantee long battery life. It needs to perform at least on the level of a Dothan 600MHz before I'm interested - web surfing is already a pain at that level of performance.

Why is it...

[religious freak]

that some /.ers seem to need to create an enemy of conventional wisdom, even when conventional wisdom is conventional for a reason?

Yes, efficiency is good. But do you really need to smear the idea of higher processing power at the same time you're pointing out the good in low electricity consumption?

I mean... really?

Re:Why is it...

[palegray.net]

You're right. Both approaches have their place. What you're observing is the manifestation of an overriding need to prove one's superior intellect. It's a sign of poor socialization.

I, for one, welcome multiple approaches to achieving multiple goals.

Re:Why is it...

[wiredlogic]

Because conventional wisdom often isn't so much wise as it is conventional.

Conventional wisdom led to the MHz war and the foolish, marketing driven decision to double the pipeline length on the P4. This directly led to lower instructions per cycle and hence lower true performance than one would expect just going by the clock rates. The average consumer is ignorant of these sort of details and the marketing folks get paid to figure out how to exploit that ignorance.

Wisdom has no place in a world where you can get ahead with smoke and mirrors.

Re:Why is it...

[Stuntmonkey]

Quite a few people in the industry now are starting to care about power efficiency at the other end of the performance spectrum too. The Green 500 list for example tracks Megaflops per Watt data for the top 500 supercomputers. Judging from this data the Cell processor looks very good.

The reasons for caring about energy efficiency at the high end are of course very different from what ARM is trying to do, which is to maximize performance within a given battery life envelope. For large installations it has more to do with operating cost and environmental concerns.

Re:Why is it...

[the_humeister]

But, is the Intel architecture really worth maintaining? The only reason I can see keeping the current IA32 around is that there's such a huge code base, and realistically if we cut the cord now, it wouldn't be too long before we could just use emulation for the old non-portable code.

I'm really not sure that it's a good idea to keep it around just because. A more or less fresh start with more modern assumptions isn't really a bad idea. Both technology as well as usage patterns have changed drastically over the decades.

That doesn't necessarily mean that Intel should be cut out, but more that keeping processors just because isn't a great idea.

If we could realistically cut the cord? The fact that the x86 ISA has persisted for so long is precisely due to the fact that we can't realistically cut the cord! Many architectures have tried for the mainstream (SPARC, PPC, Alpha, PA-RISC [were they ever mainstream?], Itanium, etc.). And now they're all dead or shoved into servers. Only PowerPC processors have been as successful as x86 chips at being placed in low-end, high-end computing applications and everything in between. Even so, with Apple out of the picture, you'd be hard pressed to find anyone selling new PowerPC-based computers geared for the general public.

Also, ISAs aren't kept "just because." If you think x86 is bad, consider that the most modern IBM Power Systems are still binary compatible with code written for machines 40 years ago!

Re:Why is it...

[smallfries]

If we didn't use the x86 instruction set we'd have to invent something quite similar to it. There are several layers of cruft in there: 2-op instructions and 100s of rarely used complex ops. But it does something that modern 3-op RISC-like ISAs fail to do - it compresses executable code. This means more sits in the instruction cache, and so less decode penalties.

The newer ISAs (Core-2 and i7) do an amazing job of extracting parallelism to boost instruction throughput. During the Mhz race years instruction latencies spiked on the P4 ISA, the more recent ISAs have really lowered those latencies to RISC-like levels.

I don't know if you've ever spent time doing low-level profiling on an Intel platform, but they do make good hardware. Here's a rule that they don't bandy around, but if your working set fits inside L3 then all of your memory accesses are free. Yes - free. I can see why they wouldn't want to make that guarantee part of the ISA, but it does mean that register pressure (the biggest problem in x86) is a phantom issue. Doing low-level paging between registers and the heap (which is what a modern compiler does with load/spills) provides a huge working set.

The article mentions efficiency - but there are plenty of situation where power consumption is irrelevant and I really do want maximum performance per thread. Any home desktop falls into this category. As any server or laptop doesn't it will be interesting to see what happens as Intel shifts from maximising performance to maximising performance per watt.

Re:Why is it...

[drinkypoo]

Well, in a word, no. The reason the Pentium IV sucked more than any other processor ever (in terms of price:performance ratio, power consumption too) is that it had an enormously long pipeline. Several pipeline stages are "drive" stages to wait for signal propagation across a chip (probably more literally to wait for inputs to stabilize.) The problem with a deep pipeline is that the cost for branch misprediction increases; not only do you have more wasted pipeline stages before your branch, but you have more potential instructions which now have to be retired from the pipeline because all other operations occurring in the same thread are now invalid. Intel went back to Pentium M which was derived from Pentium III to create the Core series of cores which we are now using.

Re:Why is it...

[pohl]

If we didn't use the x86 instruction set we'd have to invent something quite similar to it.

I agree that there would still need to be a set of primitive hardware operations. I also agree that there would need to be a way that binaries are persistently represented. But must the same thing perform both functions? Can't we decouple that with something somewhat like Transmeta's code-morphing layer? (link)

Re:fp - i win!

[koutbo6]

u mean this?
http://www.revogear.com/
These are ARM based, you can move them around, but they're no laptops.

Only 6-10?

[Scutter]

What happens when those 6-10 netbooks get sold? What about the rest of us? Seems like it's hardly worth it to build so few. They should be building them by the thousands!

Re:Only 6-10?

[Paradise Pete]

What happens when those 6-10 netbooks get sold? What about the rest of us?

"I think there is a world market for maybe five netbooks." - The ghost of Thomas Watson

Rocks don't use any electricity.

[tjstork]

Maybe we should make computers out of them. In fact, they did...

http://en.wikipedia.org/wiki/Roman_abacus

Re:Rocks don't use any electricity.

[ksattic]

http://xkcd.com/505/

MIPs?

[gzipped_tar]

The marketing term (not the architecture) MIPS == Million Instructions Per Second. It's not the plural form of some other TLA. ;)

a quarter of a watt

[Iamthecheese]

A quarter of a watt is a percentage of the static I gather walking. A processor like that is powerful enough to run a tiny GPS, an insert in my shoe. Add a little foot-pad to power a HUD and attached map and I always know where I am. This is one of many, many uses. Anyone still thinking "cell phone" is missing the point.

Re:a quarter of a watt

[Anonymous Coward]

ARMs powered by legs...

I like your idea and would like to subscribe to your newsletter.

Re:a quarter of a watt

[david.given]

A quarter of a watt is a percentage of the static I gather walking...

250mW is actually quite a lot, processor-wise. Atmel produce microprocessors that will run Linux and consume about 100mW. If you switch to a true embedded processor, Microchip's PIC24 series are 16-bit processors that will consume about 20mW at 16MHz (and less if you run them more slowly), and if you're willing to go 8-bit, you start getting into silly numbers: their PIC10 series will run (flat out) at 0.4mW and sleep at 0.0002mW. If you're used to PCs, there's a whole new world out there...

One day soon I'm hoping to see someone produce a mini laptop based around one of these 16 bit or 32 bit microcontrollers and an e-ink screen. It may not run Crysis, but it would probably run off a single AA pretty much forever, and still be useful; it would, after all, still be able to outcompute an Amiga or Atari ST...

I love ARMs...

[jonr]

They are the only chips that you can program and keep your sanity.
The ARM code is just beautyful design, one weeps with joy after struggling through x86 hell.
And computing/electric power ratio is fantastic.

Re:I love ARMs...

[Anthony_Cargile]

They are the only chips that you can program and keep your sanity.

I completely agree. The most elegant assembly I've even written, easy to optimize, and without all the legacy underpinnings of x86. Apparently the GNU folk can agree as well, because the output of any of my compiled C programs run better on an older ARM than a newer x86 chip (this is on Linux, btw).

Re:I love ARMs...

[Nutria]

They are the only chips that you can program and keep your sanity.

Your UID says you are old, but that statement indicates you are too young to have ever programmed a 68K or VAX.

Re:I love ARMs...

[Anonymous Coward]

Define too young! :P I'm 26 and spent 4 years doing 68k ASM! I even wrote a disassembler while I was in college... which earned a trip to an international teacher's convention. http://detachedsolutions.com/cmdpost/

Re:I love ARMs...

[dbIII]

Oi! Slashdot is not much more than ten years old so even a teenager might have a low UID. Also you don't have to be really old to have done stuff with the Z80 or the 6502 if you did it while you were still in school.

Re:I love ARMs...

[Henry Pate]

A lot of people that are now in their early twenties got exposed to 68k ASM with the TI-89 and TI-92 and z80 ASM with the TI-83 and TI-84 calculators.

What originally got me started programming was my TI-83 in 9th grade Algebra 2. I was horribly unprepared for the class so I learned how to make programs to do the quadratic formula, solve equations, expand polynomials and the like. Now this was just in TI-Basic but translating the math into code really helped me understand the material.

Then I found ticalc which was and probably still is the best resource for everything involving TI calculators. I must have printed almost a thousand pages of code, books, FAQs, and tutorials. I'd trace through the code to learn what I could from and then try writing something myself. Most of the games used z80 assembly and there were tons of them to look through. I think early exposure to assembly definitely improved my ability to work in higher level languages.

A few years later for Calculus I got a TI-89 which used the Motorola 68k processor, however I was never as interested in learning to program the TI-89 as I was with the TI-83. I'm sure I'm not the only one whose first exposure to programming was on the TI calculators, they probably bred a new generation of programmers through their calculators.

Re:I love ARMs...

[Darinbob]

I rather like PowerPC. Straight forward instruction set.

ARM is nice, but it's a bit compact, what with every instruction allowing an optional conditional or shift/rotate (which does give good performance with a simple design though). And the later ARM versions just got a bit cluttered with its multiple processor modes and extra instruction sets (thumb, java, dsp).

Re:Too bad the CPU isn't the only thing drawing po

[Anthony_Cargile]

Web browsers are interpreters, which are going to be slower than machines that run pre-compiled code

It's worse than that: In addition to HTML, a web browser must parse/interpret JavaScript, Java, CSS, XHTML, Flash (if Adobe ever gets onboard), and regular XML just to display the modern, JavaScript-heavy web application. This gets resource intensive if, say, using an app such as Google Docs on a netbook with little memory, since the browser keeps the DOM structure in memory, and it gets exponential if the user has multiple tabs open with an app/page in each.

A server pre-parsing HTML would mean a browser/server handshake, something IE and IIS could easily do moreso than Apache(2)/Lighttpd and Firefox/Safari/Chrome. Opera does this with their mobile platform, but it is still far from perfecting JavaScript precompilation or even delegating this to the lower-resource device at the client end.

Google was contemplating compiling JavaScript to pure native code in a story I read here on /. a while back, but how well they would maintain this for both x86 AND ARM remains another story, in addition to all of the other problems that could ensue, especially at the security level (a bug in the JS parser leading to direct remote code execution, etc.).

It's problems like these that keep 300Mhz netbooks with little RAM from being very efficient with full-scale web apps. Just my firefox I'm running now, I have about 20 tabs (mostly regular HTML) open and it runs up my dual-core CPU so high that my fan is running (not much in the background), and it eats memory like crazy. But as far as MS breaking the Wintel relationship to pursue ARM-based netbooks, I don't see it happening unless something drastic happens.

Re:Too bad the CPU isn't the only thing drawing po

[anss123]

And a 1 GHz Cortex-A8 core is probably in that ballpark.

Perhaps. A 530MHz Dothan was about twice as fast as a 600MHz Cortex A8 in a benchmark I saw. That does not mean the A8 is slower for browsing, as a browser is so complex that a simple CPU bench isn't enough. One has to sit down and use the system.

No laws overrridden

[renrutal]

From TFA:

For 30 years, the PC industry has treated Moore's Law with religious reverence. Its immutable commandment -- thou shalt double the transistors on circuits every 18 months -- created an enviable business model with consumers spurred to buy new, more powerful PCs every few years.

The actual law is about reduction of cost, not increase of performance. Other formulation says:

The transistor cost shall halve every 2 years.

ARM is not breaking any "law".

Re:No laws overrridden

[SeekerDarksteel]

To be even more precise, it's not even about cost per transistor. It's saying that the amount of transistors for which a chip will be most cost-efficient will double every two years. Moore's law could be satisfied even if transistors never shrunk in size and never decreased in marginal price if we were able to double the size of chips every two years without decreases in yield. Remember, transistors is cheap, packaging and verification is expensive.

Horsepower

[jeffkjo1]

As I read through the article (I know, I've already violated Slashdot's law, but anyway), I couldn't help but go back to this whole idea of 'under-the-hood performance.' Cars built today don't necessarily have to have the 400 cubic inch plants and 500 horsepower that they sometimes had in the 60's. Engines are half that size and half the horsepower, but because they're designed better, it doesn't matter. (Although I'd love a 500 hp engine anyway.)

As well, continuing the car analog, just because there are still some cars with 500 horsepower engines made today, it doesn't mean everyone needs one. There are plenty of tiny cars doing just fine thankyou
This article suggests that because we're not using giant oversized processors in our iPods and cellphones, that somehow we've violated Moore's law. All it really means is that putting a Ferrari engine in golfcart is pointless.

Re:Horsepower

[simcop2387]

what do you mean a 500hp engine in a golf cart is pointless? how else will i beat my ball to the green to watch it land?

Re:Horsepower

[Anne Thwacks]

Cars built today don't necessarily have to have the 400 cubic inch plants and 500 horsepower that they sometimes had in the 60's. Much as I like the idea of big-block Chevvy engines, here in Europe 3 Litres (200ci) has always been big, as we could get 100HP/litre even then, without a turbo - hell, even without fuel injection. (Hint: it pays to design decent gas flow, and we think a good engine should be balanced to do 6,000RPM without falling to bits.)

Today European/Japanese production plants regularly get over 100HP/Litre from Diesel engines with a turbo.

They think a bit differently

[tchuladdiass]

They may up the megahertz, but not at the expense of a more costly product or more power usage. Instead, the ARM chip vendors take a look at what needs the MHZ, such as video/audio decoding, and include special co-processors for those functions on the same silicon. Therefore they don't need to increase MHz for increased functionality.

It is a similar philosophy to using a script written in a slow interpreted language to drive a more complex system composed of high-speed modules written in C.

Re:They think a bit differently

[philipgar]

ARM has a couple processors already that are pretty high on the performance measurement. For instance the Arm Cortex A9 has a dual issue pipeline, and limited support for out of order processing (similar to the original Pentium processor in that regard). This chip also can contain up to 4 cores, and have up to a 2MB L2 cache. I think they can run up to about 1GHz. They also have full support for floating point and all that good stuff. I'm pretty sure ARM is also working on developing an true OoO processor that will likely be running in the GHz range which would likely be ideal for a netbook.

Remember, with a netbook, you don't gain much by lowering the CPUs power consumption to less than 5 watts or so. The reason for this is simple, the display, ram, hard drives and everything else consume enough power that it won't really help battery life very much. I can imagine though that a quad core ARM A9 at 1GHz would make for a really nice netbook. Having multiple cores is nice on those for web browsing (playing flash in the background of your tabs, etc), and also for many media tasks. It would also be great if they included a graphics chip (or gpu as part of a SoC system) that could handle h.264 decoding for the netbook.

Phil

Re:They think a bit differently

[im_thatoneguy]

It would also be great if they included a graphics chip (or gpu as part of a SoC system) that could handle h.264 decoding for the netbook.

You mean something like this:
http://www.nvidia.com/object/product_tegra_600_us.html
?

Re:They think a bit differently

[Phoghat]

I've got an Axim 51v with an ARM 624Mgh processor, running WM 5, WiFi G, Bluetooth, a VGA screen (albeit 4 inch) and a separate GPU. It fits in my pocket and I use it for reading e-books, rudimentary surfing and Word and it plays games quite well. I wouldn't mind something like this with a somewhat larger screen a little faster CPU. It wouldn't fit in my pocket of course, but that why god invented messenger bags. BTW, It's a few years old and has more processing speed and storage than quite a few of my older desktops.

Re:They think a bit differently

[Anonymous Coward]

Netbooks can use much less power than they do today. Real netbooks use solid state storage, which has negligible idle power consumption and very low read power consumption. In the future, RAM may be replaced by one of the contenders for persistent RAM (e.g. MRAM) with zero idle power consumption. Backlit TFT displays will at some point be replaced by E-Paper which only uses power to change the display. At that point, the CPU and the wireless network are about the only consumers of power, so every improvement counts.

Re:They think a bit differently

[LucidBeast]

I'm not expert on processors, but Nokia phones have ARM processors and can do h264 decoding. I don't know if there is a separate chip for that, but knowing how slow the processors is it must be somehow accelerated.

Re:They think a bit differently

[BikeHelmet]

You got modded informative because you had nVidia in a link.

TI always pairs their Cortex CPUs with beefy DSPs capable of very complex decoding. The OMAP 3530(in use in devices right now) is able to decode 720p h.264 by offloading it to the DSP. A DSP is similar to a GPU, but this one lacks floating point capabilities. It's just really fast for integer stuff.

They'll probably pair an even faster one for the Cortex A9's, enabling 1080p h.264.

Re:They think a bit differently

[shutdown -p now]

Backlit TFT displays will at some point be replaced by E-Paper which only uses power to change the display./quote.

I doubt that's gonna happen for several reasons.

First, e-Paper refresh rate is horrendous, and it is unlikely to improve to the same levels as TFT for the foreseeable future due to inherent technology limitations. Moving solid particles around (which is what eInk does) is always going to be slower.

Second, while eInk screen doesn't have to be redrawn all the time, when you actually need to redraw it, it may well use more power (I'm not sure about that one, actually, but it would make sense). And how often would you need to redraw? Well, if you leave the mouse in, every time the pointer moves... so you'll need a touchscreen and no pointer. And as few smooth animations as possible. In fact, it may well require a total redesign of UI to make it work - so forget existing apps.

Third, eInk is useless in below-average lighting conditions. Remember that it doesn't emit light by itself, and it is not transparent, so you can't use a backlight (and if you did, that would suck up power just as it does in TFT). And you can't make the particles transparent, because the whole point of technology is to make them reflect light...

Fourth, eInk color gamut isn't going to be any better than printed stuff, ever, for obvious reasons. This may be good enough for some stuff (most office/productivity apps, web browsing), but forget about decent video (remember last time you've read a movie review in a journal? remember what the screens looked like... that's right, crap).

All in all, I think that it might work out, but only for a true "netbook" - a device that's only useful for surfing the Web, and nothing else; and even then with some interactive stuff (e.g. YouTube) crippled. Also, given the pace of technological advances in e-paper, it's going to take a decade at least before we get that stuff in production (there's still no working color e-paper available for use in production e-reader models, even though the first prototypes were shown over 2 years ago).

monster market

[philospher]

I think the ARM netbooks are going to have a monster market, like eventually over 100 million a year.

That may sound crazy, but you have to look at the demographics. There are about 6 1/2 billion people in the world. About 1 1/2 billion are in the developed world or the richer parts of the developed world. They all have computers. At the other end are about a billion who are are desperately poor.

That leaves around 3 billion who are in-between. These are the people who have enough money to buy things like bicycles, motor bikes, televisions, and cell phones. A great many would love to own a computer, and indeed many of them spend a lot of time at cybercafes. But they can't afford the price. And there is another problem, namely that half of these people live in areas with no electricity, and for most of the rest the electric service is very eratic.

The first generation of netbooks was too expensive for this gigantic potential market, and besides they used too much electricity. But the new ARM netbooks will be enough cheaper for perhaps 500 million more people, and they will use far less electricity, too. Furthermore prices are just going to keep going down. Pixel Qi is planing on designing $75 models in a few years. Every time prices drop another huge group will join the market.

This all is a huge problem for Microsoft. On the one hand, it would hate to charge the very low license fees it would need to get anywhere in this new market, on the other hand it can hardly afford to ignore it.

Incomplete quote

[MeanMF]

That last quote in the summary should read "...6-10 ARM-based netbooks running Linux and costing just around $200 should arrive this year starting in July and be done booting up sometime in early August"

1982?!!??!

[mcrbids]

...your average phone is more powerful than your average computer was in 1982.

Sonny, I was THERE in 1982 and I can tell you that my phone (an HTC Mogul) with its dual-core 400 Mhz ARM CPU knocks the socks off the 386 I had aound 10years later, around 1992! In fact, I can run DOSbox and run all the same games I used to play on my fire-breathing 386DX25 in emulation !!

If my phone today was released in 1982 it would probably have been considered a controlled military tool and banned from use by nonmilitary personnel!

Psssssttttt! Wanna guess what I'm typing this post with?

Re:1982?!!??!

[rackserverdeals]

Psssssttttt! Wanna guess what I'm typing this post with?

Your fingers.

Re:1982?!!??!

[Anonymous Coward]

Aw, how innocent you are.

Re:1982?!!??!

[Jah-Wren Ryel]

Psssssttttt! Wanna guess what I'm typing this post with?

Your arm?

Re:1982?!!??!

[tzot]

Shame I already used up my mod points...

Misunderstanding Moore's Law.

[Anonymous Coward]

Moore's law isn't about what shall be done, or what should be done. It's about what can be done.

The size of a transistor on silicon has been steadily shrinking ever since the introduction of the first silicon chip. That's been driving down prices, and raising efficiencies. For example: the 6502 drew up to 160 mA at 7V. That's a little over one watt of power. The most power hungry Intel CPU on the market draws about 150 times that amount, but can do well over 150 times the work - getting hard facts is difficult, but I'd suggest four or five orders of magnitude, if not more.

So with the shrinking transistor, you can do three things. You can make the CPU more power efficient - able to do the same amount of work with less power. Or you can make the CPU more powerful - able to do more work, for the same amount of power. Or you can do both - able to do a bit more work, for a bit less power.

Intel has chosen to make their CPUs more powerful, at the cost of keeping the power usage high. ARM has chosen to make their CPUs a little more powerful, for a bit less power. Both are equally valid paths.

In the long run? Both choices will carve out their niche in the market place. There'll always be room for computing power at any price. But for the typical Joe Blow off the street, the ARM tradeoff - less performance, at a lower price - is more likely to be useful, assuming the software is there ... and with Linux, it pretty much is there.

Think "co-processor"

[msobkow]

ARM chips are famous for including special instructions and supporting silicon for things like MPEG4 encoding/decoding, MP3 encode/decode, etc. The "main" CPU core isn't involved in these "streaming" instructions, just the parameter setups for them. Given enough "heavy CPU" workloads implemented as custom silicon, the main CPU on an ARM chip can be relatively idle as all the heavy lifting is done by the stream coprocessors.

Look no further than the Nvidia Tegra

[CAIMLAS]

It would appear that you're really not aware of what's out there on the ARM department right now. Marvell is not the end-all, be-all for ARM processors, and the (relatively ancient) StrongARM CPUs are not even remotely comparable to what's on the market, in terms of performance.

Look at the Nvidia Tegra for a perfect example of ARM walking all over Atoms - per clock, per watt, and per actual performance.

There are a handful of other notable ARM chips out there right now which, while not comparable to the Tegra directly, offer considerable options above and beyond the Atom. Snapdragon and Tegra are just two examples; there are many others.

The performance is there, and has been there for quite a while. ARM chips do a LOT of things which an Atom couldn't come close to doing effectively (that fanless set-top box that does digital to analog conversion, or the DirectTV dvr, for instance).

Windows for ARM?

[GreatDrok]

OK, so WinCE/Mobile/whatever the hell isn't really Windows. It won't run all your apps. Linux won't either but is much more functional than Windows Mobile. Where will this leave MS with their strategy of forcing companies to bundle Windows instead of Linux on their Netbooks? What about the next OLPC which isn't supposed to have an Intel compatible processor either? Is this all a strategy to spoil MS's fun? I sure hope so!

Re:Windows for ARM?

[FrostDust]

Why do you think Windows has woken up to the "cloud computing" craze with their Live brand of services? All Microsoft has to do is get a decent version of Internet Explorer running on Windows Mobile, and maybe tweak WM to expect a bit more generous system resources (netbooks vs. smartphones).

Thanks to everyone trying to make their programs server side nowadays, it doesn't matter what processor you're using; just as long as you have a net connection you can do anything you'd expect to do on an x86-based netbook.

In joke on page 8 of the PDF

[Hal_Porter]

Look at this PDF, page 8, top left picture

It's actually from here

http://www.snopes.com/inboxer/hoaxes/computer.asp

That said, I suspect whoever wrote it was aware of the Snopes article.

Re:fp - i win!

[the_humeister]

I know it's not ARM, but this thing's also not x86 - a MIPS-based mini-notebook: Alpha 400 MIPS netbook.

They're pretty inexpensive too. I might pick one up just to play around with it.

Re:Still focus on a single metric

[thaig]

NVidia's Tegra is sort of proof that you don't need to worry too much about this.

With all sorts of things built in e.g. powervr 3d, dsp etc, and with multiple cores, ARM CPUs can be quite powerful - enough to do a lot of cool stuff. I should think that consumer-type video editing would be no problem on some of the recent CPUs. A lot of arm SOCs are designed for phones with video cameras and they have hardware assisted compression/decompression. They are starting to have enough RAM too, although that uses more power.

What will be nice will be that you can use the thing the whole day and take it with you everywhere and not worry about finding places to recharge it or about carrying the transformers etc.

You can still do your hardcore 3D rendering at home if that's what interests you, because the cost of your device will be low enough to allow you to still buy a stonking-great, deskbound number cruncher.

Re:Too bad the CPU isn't the only thing drawing po

[Quantumstate]

If you do try this be sure to use a virtual machine or a real low powered machine. Web browsers adjust their memory usage based on what system they run on and with other factors like cpu speed it is very difficult to extrapolate from a fast machine to a slow one.

Re:Too bad the CPU isn't the only thing drawing po

[jabuzz]

A ARM Cortex A9 dual core clocked at 1GHz, that is a top of the line ARM, would run rings around any Atom while consuming a fraction of the power.

The trick is that the ARM instruction set is *WAY* more efficient than the x86. The fact that the current ARM's are basically in order units is less important due to the design of the instruction set.

You are right about the power though. The ARM needs to be coupled to a low power chipset, but guess what these also exist as well.

It also needs a low power display. Now if I could just get a netbook with an Cortex A9 a GB of RAM, with 8-16GB of flash and a LCD from a OLPC XO-1 for 200 USD I would be well chuffed. I would expect such a netbook to have around 10-12 hours battery life.

Re:Too bad the CPU isn't the only thing drawing po

[Com2Kid]

The trick is that the ARM instruction set is *WAY* more efficient than the x86. The fact that the current ARM's are basically in order units is less important due to the design of the instruction set.

A minor correction here: The ARM instruction set is simpler and is much less feature rich than x86. This, combined with self imposed limitations (mostly around in-order execution) ARM (the company) is able to design CPUs with a much lower transistor count than x86 chip designers are capable of managing.

Not having a huge instruction decoder, having to do instruction reordering, or basically doing any of the things that makes x86 so damn fast, and staying a generation or two behind on manufacturing techniques to avoid leakage issues, enables ARM CPUs to have their amazing power profile.

The instruction set is kind of a mess really. It has been hacked onto a number of times, with the latest additions really showing signs of having wedged into the existing instruction encoding space. With features like the original (crappy) thumb, and now the fixed "Thumb2" (which we are all supposed to be calling Thumb and ignore the old Thumb, or something like that), and the fark-up that is ARM's floating point support (They have 3 implementations, 2 of which are still in use, and those two versions respond dramatically different to some basic key floating point operations), the ARM instruction set isn't nice per say, but ARM did the right thing and by keeping their eye on power consumption always.

(For those who are getting linguistically confused: ARM the name of the company, the name of their CPU line AND the name of their instruction set. Oh it is also the name of their reference manual, the ARM-ARM.)

200?

[nurb432]

We have heard that before, and the 'super cheap' never quite pans out and ends up 2x.

Re:Too bad the CPU isn't the only thing drawing po

[SpinyNorman]

One motivation for putting an ARM in a netbook is to make a better product (overall coat-performance-battery life trade-off), not just a cheaper one, so why would a manufacturer not put a decent/large amount of RAM in one? Have you checked RAM prices recently - it's practically being given away.

As far as web compatibility, note that the iPhone is ARM based and has a decent browser (youtube compatible since youtube switched to H.264 video), and incidently Adobe is trying to get Flash working on the iPhone...

FYI Google's new JavaScript VM is here in the Google Chrome browser... The JVM is called V8 and does indeed compile to native code when it wants to (JIT), and runs rings around other JavaScript implementations (such as that in Firefox) in terms of speed. Considering how simple ARM machine code is (it's a totally orthogonal instruction set)it's hard to imagine porting the JIT compiler to ARM would be a big deal. As far as security, JIT makes no difference.

Re:Still focus on a single metric

[zsau]

Insightful? Beating a dead horse more like it. The sort of computer these things are going in will be in a separate category of devices than your workstation. Want a superfast box for graphical modelling? Buy a workstation. Want a lightweight, low consumption computer to access the internet at a cafe? Buy a netbook.

It's already like that. You'll find most big-enough computer companies are happy to sell you servers and workstations and desktops and laptops. They'll even give you specs if you know enough to know what's important in your workstation. We're just adding another category. It's not the end of the world, and it's not the end of graphical modelling.

Although, to the extent your point is that the article was an overexaggeration, well, it's slashdot. What do you expect? I'm going to have to stop visiting it again...

PSP 2 and PSP 3 still run at the same clock speed

[tepples]

PSP (not the most relevant example) might be 300 MHz now, what about PSPII, still 300MHz? Doubtful.

The first PSP ran at 222 MHz. (The 3.50 firmware upgrade unlocked 333 MHz with the WLAN off in games that require firmware 3.50 or later.) The PSP-2000 added RAM, but not much else in the sense of processing power. The PSP-3000 didn't add processing power either.

Obligatory TA reference

[OneAhead]

The battle between the Core and the Arm is about to begin! Finally I can put my stock of tins of baked beans to use!

is ARM licensing terrible?

[DrDitto]

I don't remember the details from a group of computer architecture friends interested in forming a startup, but if you are interesting in licensing the ARM instruction set to develop your own ARM processor for sale, good luck with that. If I recall correctly, you get something like 12 months of exclusive rights, but then you have to give up your design to ARM. Someone correct me if I am way off here. Someone correct me if I am way off base.

Uhm, Intel makes ARM chips too

[BitZtream]

Its going to be hard for ARM to kill intel. Thats roughly the same as saying x86 or itanium is going to kill intel.

Generally when people are buying a product your company makes, your company does better, not worse.

Guess no ones heard of XScale? What am I missing here?

baths are bad???

[Nivag064]

Actually, baths are bad. Specifically if you wash your skin, you remove the protective layer of oil - this is a fact.

Recently, a few years ago, they paid a bunch of students not to wash for a few months. They monitored them quite closely. After about 6 weeks or so, the skin stabilized in terms of pH etc..

In a "confrontation" about the time Malaysia was formed, British and New Zealand soldiers made many patrols. It was found that the NZers had a higher percentage of their soldiers reporting sick, but not because they wanted to avoid going on patrol (British officers had very high regards for the courage and endurance of the NZers). It was put down to New Zealanders generally being used to cleaner environments than the British soldiers.

-Gavin