ARM, Intel Battle Heats Up 260
An anonymous reader writes "Low-power processor maker ARM Holdings is stepping up rhetoric against chip rival Intel, saying it expects to take more of Intel's market share than Intel can take from them. With Intel being the No. 1 supplier of notebook PC processors, and ARM technology almost ubiquitously powering smartphones, the two companies are facing off as they both push into the other's market space. 'It's going to be quite hard for Intel to be much more than just one of several players,' ARMs CEO said of Intel."
Simple math, silly! (Score:1, Interesting)
Of course ARM will take more from Intel than Intel will from ARM. But it's stupid math.
Take the desktop market. If ARM has 0% of the desktop market, and Intel has 65%, there's nothing for Intel to take from ARM. If ARM sells just 1 desktop, they've taken more from Intel than Intel took from them. But it's a meaningless stat.
Random idea (Score:5, Interesting)
Intel gets my respect for being one of the few companies to invest heavily into research. Seriously, they do a lot of "fundamental research" work, and so far, it's worked well for them. They develop products all the time that never get released because they're too "experimental" - Larrabee is the example that comes to mind first - and justify the expense because the information learned is worth the $100M they spent on an unreleasable product.
Intel, you can hedge your bets. Take one of your teams - rumor has it the Itanium team won't be working on that much longer - and tell them to make a desktop-quality ARM processor. You've already got the ARM license, do something with it. Figure out how to bump up the clockspeed (if *Apple* can do it, so can you), throw cache at it, bring the core count up to eight or so. Target your own Core i3 chips both in speed and in cost.
You do that, Intel, and you basically can't lose (barring sudden inexplicable incompetence). If the ARM desktop project completely fails, well, you just proved that x86 chips are unbeatable on the desktop market (which will never completely disappear). If the project succeeds, you'll win no matter which architecture comes out on top, and you'll have the advantage of having an experienced ARM team to help you take the best features of ARM and put them in your mobile x86 chips.
Re:Random idea (Score:1, Interesting)
*Apple* doesn't do shit, 99.99% of their processors (development included) is from Samsung Korean shop
Why would they want to do that? (Score:4, Interesting)
x86 is a big advantage for Intel. Not a lot of people have a license to use it, and the ones that do either don't do a very good job (AMD) or haven't done anything yet (nVidia). It also gives them binary compatibility with so much out there. It is big to be able to run a bunch of programs with no recompiling (and even with recompiling an architecture switch can be a pain).
Were Intel to do an ARM chip like that, it would be an internal hedge, you wouldn't see it unless there was a reason. There would be no reason to sell thing thing and give the ARM market credibility against the x86 market. They'd only introduce it were it clear ARM was the way they needed to go.
There's heavy inertia on x86 as well and it may never change. Really modern compilers and microarchitectures have rendered a lot of the old school RISC vs CISC and arguments like that moot. Turns out you can use pretty much whatever ISA you like and make the chip work well, and CISC isn't a big deal for modern compilers.
If you see Intel do ARM chips, I think it will just be for mobile phones and tablets. If their attempt to muscle in to that market with x86 chips is unsuccessful, they may elect to play the ARM game, which they'd have an advantage on most other fab since they are generally a node ahead in process technology.
I can only see a desktop ARM chip getting released if ARM starts to become the One True Way(tm) and I don't think that is all that likley.
Re:Where are the products ARM? (Score:5, Interesting)
The only Intel consumer product ARM licensees are currently able to threaten is the Atom product line. Apart from that, both kinds of CPUs are simply serving two completely different purposes.
Yeah, but...
The only reason my ultraportable laptop has an Intel Celeron U3600 in it (1.2GHz dual core arrandale, 18W TDP) instead of an ARM is because I couldn't find a laptop in the same class with an ARM chip. They're serving completely different markets, but ARM is easily powerful enough for most users (just look at the R-Pi running 1080p H.264 video over HDMI), and there's absolutely no reason my laptop needs an x86 processor. I just couldn't, at the time, find a 13" ultraportable with an ARM chip in it. (closest I could find was an ASUS Transformer, but I want to run a full desktop OS on it, not Android, and it was actually more expensive than I paid for my laptop).
BTW, if somebody can find one now, I'd love to hear about it... I'm not in the market right now, but I like to know about that kind of thing for when I'm shopping next time... and also so I can make suggestions for family. :)
Re:Random idea (Score:5, Interesting)
You want Intel to make an ARM chip that is competitive with x86. Intel will never, ever do that if they can possibly avoid it.
Intel dominates in x86, and they make good profits on x86 chips. As noted in TFA, Intel would be just another ARM source out of many, and they would make less on an ARM than on x86. nVidia, on the other hand, is no longer friendly with Intel and has no reason not to build a super ARM as you would like; and in fact they seem to be working on it. Google search for "Project Denver".
The first Project Denver silicon is rumored to be 8 ARM cores, 64-bit, with a 256 CUDA core GPU on the same die. I would love a smartbook with that chip, but I think they will be able to sell that as a blade server platform as well.
steveha
Sounds like a buyout is in the works... (Score:1, Interesting)
Re:And Intel has a trick up their sleeve (Score:4, Interesting)
"ARM have a simpler, more efficient architecture"
ARM fans say that like an article of faith, but I never see any proof of it. Show me the proof that an ARM chip can do more than an Intel chip, controlling for all variables. Take something like, say, solving systems of linear equations, one of the grand daddy of computation tests (linpack is a popular tool for it). Show me an arm chip doing better on linpack in terms of Mflops/FPU area or something, then I'll say ok it is more efficient.
However it seems to me all ARM fans have to go on is that ARM makes tiny chips. Ok, they do... so what? Intel's chips are bigger but they do a hell of a lot more. Not only do they do faster calculations, but they have more features (like a 64-bit architecture). So calling them "inefficient" is silly without some kind of metric.
So like I said, how about FPU area? ARM chips aren't 64-bit, but their FPU should be, and I've seen ARM chips with vector units. So take the area of the FPU on chip, and see what it can crank away on linpack. Then take the area of an Ivy Bridge and do the same. Adjust for clock differences and see what you have.
you sound like an industry outsider (Score:4, Interesting)
I have Cortex-A15s in the lab right now, finished simulation and verification, and samples are being boxed up this week, and we're shipping them this year. And for quantity we can do as many as TSMC can make for us. Then it boils down to if any devices makers are willing to pay the premium we have to charge for the bigger faster chip. You will see about half of Jellybean devices running a Cortex-A15, and nearly all Windows 8 devices when it first ships.
Re:Where are the products ARM? (Score:4, Interesting)
http://www.calxeda.com/ [calxeda.com]
HP will be using them in products.
The video I saw talked about replacing 20 racks with one half rack. Not for all supercomputing tasks of course, but for web serving or Hadroop it works.
IIRC they were using four core SoCs with built in fabric. Obviously the same approach will work with A15 (and 64-bit when that come goes into production)
Windows on ARM is not about servers, but the same solution for laptops will work in low-energy servers.
Fantastic Sophie Wilson quote (Score:5, Interesting)
See the source. [reghardware.com]
Today one of the most significant features of the ARM family is its low power consumption. But that hadn't been an initial goal, according to Furber. “We designed the ARM for an Acorn desktop product, where power isn't of primary importance. But it had to be cheap. Cheap meant it had to go in a plastic package, plastic packages have a fairly high thermal resistance, so we had to bring it in under 1W.”
The power test tools they were using were unreliable and approximate, but good enough to ensure this rule of thumb power requirement. When the first test chips came back from the lab on the 26 April 1985, Furber plugged one into a development board, and was happy to see it working perfectly first time.
Deeply puzzling, though, was the reading on the multimeter connected in series with the power supply. The needle was at zero: the processor seemed to be consuming no power whatsoever.
As Wilson tells it: “The development board [we] plugged the chip into had a fault: there was no current being sent down the power supply lines at all. The processor was actually running on leakage from the logic circuits. So the low-power big thing that the ARM is most valued for today, the reason that it's on all your mobile phones, was a complete accident."
Re:And Intel has a trick up their sleeve (Score:2, Interesting)
Intel is, as usual, a node ahead of basically everyone else.
And they need it with their architectural disadvantage.
Other fabs just got their 28nm half node online not long ago, late last year.
Half node is the new full node. TSMC has a good chance of having 20 nm before Intel has 14 nm, so there is the possibility of a window of up to a year when TSMC is on 20 nm with intel at 22 nm. Interesting game.