11 Comments

Summary:

Notebook makers are reportedly bidding on chip supplies from both Intel and those provided by vendors using the ARM architecture, presumably to compete better on pricing with Apple. The real story is that the next round of chip wars between Intel and ARM licensees is here.

MOD-40924_Tegra3_Chip

Notebook makers are bidding on chip supplies from both Intel and those provided by vendors using the ARM architecture. DigiTimes, which tracks Asian hardware suppliers and device makers, reports that the reason companies are bidding on both chip types is to better compete on price with Apple. That may be true, but in the long run, it’s not the real story.

More important is that this action is the next round of the chip wars between Intel and those who build ARM-based CPUs.

This battle has been building for several years as mobile computing — on smartphones and tablets — has quickly matured, while traditional desktop computing has stagnated. As early as last December, for example, smartphones began outselling personal computers and those sales show no signs of stopping. The biggest difference, at least in terms of the chips that power these two markets, has been one of strategy.

Until recent years, Intel focused its efforts on what’s called the “clock speed” of CPUs, rapidly increasing the performance of computer chips to better handle desktop operating systems and processor-intensive applications. Less thought was given to reducing the power consumption requirements of these chips.

Contrast that to chips built on the ARM architecture, which is licensed out to chip-makers such as Nvidia, Qualcomm, Texas Instruments, Freescale and a host of others. Instead of the “top-down” strategy of boosting performance first and focusing on power requirements second, ARM chips have used a “bottom up” approach. Early ARM chips weren’t capable of running complex software, but could run for days between charges. Once the power requirements of the silicon were effectively managed, ARM chips began to ramp up performance; most recently with quad-core chips that can offer 16 hours of high-definition playback on a tablet.

These chips now power everything from phones to tablets to e-readers, all markets that have two things in common: They’re all growing, and Intel’s silicon is inside few of them. This confirms early analyst expectations from January 2010 when ABI predicted that more mobile devices would be running on ARM chips instead of Intel’s silicon by 2013.

As if that wasn’t bad enough for Intel, Microsoft demonstrated in January that its next version of Windows, possibly available in 2012, doesn’t require Intel’s chips to run. Instead, Microsoft is optimizing its operating system for use on the battery-friendly ARM-based chips; now that these CPUs have improved their performance, they’re capable of running Windows on notebook computers. Microsoft’s software will still support x86 chips from Intel, likely for many years to come, but the trend is clear.

Intel continues to shift its strategies towards improving battery life with its chips, reportedly planning to speed up the product cycle. Plus it’s trying to jump-start what it calls the “ultrabook” category of notebooks. But the damage is already done, especially if notebook makers are already bidding on chips from both Intel and from ARM licensees. At least Intel doesn’t have to worry about the silicon it makes for large servers… Oh wait; that’s the next battle after this one, which is also ramping up soon.

  1. Michael Hunter Friday, November 11, 2011

    A simple “amount of silicon sold” argument isn’t going to get it done. This market is currently in a lot of flux. The spread in margins is large between chips and makers. And the market is growing. It might be that Intel takes it in the shorts. It might be that ARM licensees only sell lower margin embedded parts. It might the market grows and both prosper.

    Share
  2. Intel should be scared of this simple fact: FAR more people in the world own phones than they do computers.

    Share
    1. I don’t think it matters eventually. Whether a TV, mixing console, robot, surveillance system, server farm or Dick Tracy watch, everybody is going low power and Intel has always been one day, one step and several watts TDP short. Jobs, in his bio, is quoted as finding the Intel organization to be slow to change and the chips just did not offer the SoC efficiencies needed. Forget die size, Intel is always a step behind. The biggest question on the ARM side is whether multiple cores or multiple processors can ever be effectively harnessed by software or whether the potential will be frittered away.

      Why power efficiency in all things? It isn’t just about the energy bill in dollars and cents. Computing now commandeers a very large chunk of our energy production and if computing becomes the ubiquitous fabric of the world’s life, the total in 10-25 years becomes just that more hideous.

      A sidebar to all of this – isn’t ARM RISC-based and x86 CISC-based? I know the differences are far less than they were in the past, but it would be interesting if circumstances regarding power use have shifted, in such a way, that RISC has come back to exact revenge.

      Share
      1. Intel is still 2+ years ahead of all when it comes to manufacturing. A lead that will be near impossible to overcome. The Si design/RTL efficiency and Power lead that ARM based designs currently have over Intel have been getting smaller and smaller over the last couple of years. In the end this is a manufacuting and margins game.

        Share
      2. ARM was significantly influenced by the Berkeley RISC project, and is usually classified as a RISC architectures, though it has some oddball features.

        x86 is probably more accurately classified as something that is not RISC. It’s essentially the result of evolving a 4-bit CPU (Intel 4004) over a long time. It’s got all kinds of warts because of this. The true CISCs are things like the DEC VAX and Motorola 68K and the Intel iAPX432, which were explicitly designed to “reduce the semantic gap” between machine code and high level computer languages.

        Share
  3. ARM have a long 12 monthsahead of supplying laptops without suitable OS.How popular will that be? When the popular OS, Windows8, arrives it will not run legacy software. How popular will that be? ARM netbooks failed before because they didn’t have Windows. What has changed? With luck Intel will keep their prices so high that nobody buys the Intel laptops.

    Share
    1. Try re-reading the article.

      Share
      1. Why don’t you? In 12 months ARM will begin to have (increasingly) marketable laptops, not before. How many people will buy an Ubuntu laptop?

        Share
  4. A minor quibble: “quad-core chips that can offer 16 hours of high-definition playback on a tablet” is actually a product of CPU + battery pack, not CPU alone.

    Other than that, thanks for this update.

    Share
  5. Sound like a shallow analysis.. First, what is the source of the chart shown in this article? Second, what platforms are shown on this chart? ARM has almost 0% of the PC business (not according to this chart that is showing them at 40% as of 2011). It sounds like an article similar to those who believe the PC is dead.

    Share
  6. ARM vs Intel: Just starting or already over? http://t.co/8PFE17Zf

    Share
  7. ARM is the dominant CPU architecture in all mobile devices- that battle has been over for a long time. ARM is making great inroads into automotive, a long battle that they are winning. Now the shift to cloud based computing reduces INTEL’s architectural legacy advantage. Chromebooks don’t care what the CPU is. Finally, INTEL is being threatened on inside the last defense of servers. ARM based servers are coming from HP. INTEL has great manufacturing, but that won’t help much when they are literally surrounded and under siege by every other semiconductor manufacturer.

    Share

Comments have been disabled for this post