ChipDesign
Latest
Amazon eyeing up TI's smartphone chip business, according to Israeli newspaper
Remember when Texas Instruments revealed it was planning to dump its mobile processor business in favor of embedded systems? Israeli business sheet Calcalist is reporting that Amazon is in "advanced negotiations" to snap up that part of TI's OMAP division, which currently supplies processors for the Kindle Fire and the Nook HD. The paper suggests the company is emulating Apple's purchases of chip designers in order to lower the price of future hardware -- which it currently sells at cost.
Low-power chip guru quits Samsung for Apple, with heavily implied implications
The iPhone 5 already proved Apple's desire to move away from existing processor designs and exert more control over these fundamental components. Is it too crazy to imagine that Cupertino would like the same sense of freedom with its laptops? Perhaps not, especially since the biggest company in the world just hired a guy called Jim Mergard, who helped to pioneer AMD's low-power Brazos netbook chips and who had only recently moved to Samsung. A former colleague of Mergard's, Patrick Moorhead, told the WSJ that he would be "very capable of pulling together internal and external resources to do a PC processor for Apple" -- possibly based on a mobile-style SoC (system-on-chip) rather than a traditional PC approach. That's pure speculation of course, but funnily enough it's where Intel seems to be headed too.
How the iPhone 5 got its 'insanely great' A6 processor
It's hard not to be impressed by the A6 engine in the new iPhone 5, since it's now proven to deliver a double-shot of great performance and class-leading battery life. But silicon stories like that don't happen over night or even over the course of a year -- in fact, analyst Linley Gwennap has traced the origins of the A6 all the way back to 2008, when Steve Jobs purchased processor design company P.A. Semi and set one of its teams to work on creating something "insanely great" for mobile devices. Although Apple is steadfastly secretive about its components, Gwennap's history of the A6 (linked below) is both plausible and a straight-up good read for anyone interested in the more fundamental aspects of their gadgets. Whereas the A5 processor stuck closely to ARM's Cortex-A9 design, Gwennap is convinced -- just like Anandtech is --that the A6 treads a very different path: it's still based on ARM's architecture and it's likely fabricated by Samsung using a cutting-edge 32nm process, but it's an in-house vision of what a mobile chip should be. It's the culmination of four years of hard work and perhaps half a billion dollars of investment. That's not to say it's the most powerful chip out there, or even the chip most tailored to its host device -- after all, Samsung also designs great chips for some of its own smartphones. Indeed, Gwennap says that the A6 is probably a dual-core processor that is no more complex than Qualcomm's Snapdragon S4 (let alone the S4 Pro) or the forthcoming generation of Cortex-A15 chips, while its clock speed could be as low as 1.2GHz -- versus a 1.6GHz quad-core Exynos in the Note II and even a 2GHz Intel chip in Motorola's new RAZR i. However, Gwennap predicted that even if the A6 falls short of its rivals "in raw CPU performance," it'd make up for it in terms of low power consumption -- which is precisely what we've confirmed in our review.
AMD teases next-gen Steamroller CPU, aims to Bulldoze the competition
AMD's Bulldozer CPUs remind us of Betamax (or MiniDisc), in that its superlative design hasn't been embraced thanks to one or two humbling limitations. However, Mark Papermaster, Sunnyvale's new CTO, took to the stage at Hot Chips to show how he's changing the situation with the third-generation Steamroller architecture. It's rowing back on the more experimental elements of the design, scrapping the single shared fetch-and-decode hardware in favor of dual-cores that should double the amount of instructions it can handle. It's hoping to make performance gains of 15 percent, clawing back some of Intel's lead, and is considering roping in the GPU to help with the heavy-hitting in future versions. The chips will be built at Globalfoundries 28-nanometer line and are hoped to be out at some point next year.
China Times: HTC wants to develop its own processors for low-end phones
Encroaching into the semiconductor business might not seem the most obvious move for a phone manufacturer that's trying to unify its efforts. Nevertheless, China Times reports that HTC has signed a "memorandum of cooperation" with ST-Ericsson to co-develop a new dedicated chip for low-end handsets coming out next year. Since ST-Ericsson is a fabless chip designer, HTC won't risk getting silicon between its fingernails. Instead, if this agreement is what it seems, the Taiwanese manufacturer may simply want more direct control over its supply chains and to reduce its current reliance on ready-made designs from Qualcomm or NVIDIA. After all, it can't be easy for HTC's new CFO, looking on while others gobble up those margins.
Fuzhou Rockchip hypes RK2918 chip for bargain ICS phones and tablets
Could Fuzhou Rockchip's new RK2918 be the next go-to processor for a budget Android 4.0 experience? We already spotted this fleck of silicon running in the company's PAD prototype (shown above) and now we're told it'll also power a China-targeted ICS smartphone made by AirTouch, as well as an 'eHome Cloud Solution' that'll somehow make it easier for smartphones, tablets and PCs to share resources over a home network. Rockchip claims it's becoming a "virtual behemoth in North America," but only hands-on time with its latest products will tell whether that means 'big' or just 'monstrous'. Read on for more ambiguous details in the press release.
Koomey's law heckles Moore's in the post-PC world
Around the same time most years, (2007, 2009, 2010), someone heralds the death of Moore's law. This time it's Stanford University's Dr. Jonathan Koomey, who has found that energy efficiency roughly doubles every two years. With the rise of mobile devices, we care less if our phones and tablets can outpace a desktop and more about if a full charge will last the duration of our commute -- reducing the importance of Moore's law. Historically, efficiency has been a secondary concern as manufacturers built ever faster CPUs, but Koomey believes there is enormous room for improvement. In 1985, Dr. Richard Feynman calculated an efficiency upper limit of Factor 100 Billion -- since then we've only managed to achieve Factor 40,000. Let's just hope Quantum Computing goes mainstream before next autumn so we can get on with more important things.
Intel researching "carbon nanotubes" for chip design
While Moore's Law has held up pretty well over the last 40 years, it may not be able to stay true forever. It turns out that as the components inside semiconductors get smaller and smaller, electrical resistance goes up, thereby reducing performance; experts say that eventually there will be a breaking point for "copper interconnects," reaching the point where Moore's Law falls apart. Scientists have been well aware of this roadblock, and have invested heavily in everything from quantum computing to optical processors. Intel is also working on a solution for this electrical engineering problem by attempting to determine whether these semiconductor interconnects can be replaced by carbon nanotubes. The ubiquitously researched microscopic tubes can conduct electricity far better than metals, due to their "ballistic conductivity," a property where no electrons are dispersed or blocked. But, the problem with carbon nanotubes, as CNET reports, is that they're really tough to mass produce; once created, some act as great semiconductors, while others don't. So now, Intel has to figure out how to get carbon nanotubes to act more uniformly, or to separate the bad ones from the good. Thankfully, consumers won't have to worry about this problem for about another decade, which is why Intel has brainiacs working on a solution as we speak.[Via Slashdot]
