Intel's Developer Forum (IDF) starts on Tuesday, and the chip manufacturer is expected to discuss its Ivy Bridge refresh, Haswell, talk about its upcoming 32nm tablet platform, Clover Trail, and possibly share a few details on its 22nm Atom SOC's. Performance per watt is going to be a major focus at IDF this year; Haswell is rumored to be capable of idling below 10W. Intel always uses IDF as a demonstration platform and roadmap update, but it's particularly important this year, when competing foundries have struggled with 28nm and Intel itself just cut its Q3 guidance.
Several of Intel's rivals are gearing up their own demonstrations for this week. In AMD's case, that's normal -- Sunnyvale almost always does its own event during IDF -- but ARM and Apple are newcomers to this particular fray. Here's what to expect from each:
We expect AMD's
event to focus on two things: The company's SeaMicro
acquisition and design wins around Trinity's 'ultra-light' form factor. AMD will probably demo its upcoming Opteron 6300 chips; the new, Piledriver-based cores are expected to offer significantly improved power consumption and higher clock speeds compared to Interlagos. An extra 200-300MHz at the same TDP, combined with a small IPC boost, would significantly improve Piledriver's ability to compete in server workloads.
Sunnyvale is likely to focus as much discussion on SeaMicro's "Freedom Fabric" than on any discussion of processor performance. According to the company's published whitepapers, SeaMicro's fabric is designed to virtualize a great deal of capability that's normally implemented in hardware. The 2010 document claims that SeaMicro's designs can reduce server power consumption by up to 75% and takes up 1/4 of the space as a conventional server deployment of comparable power. AMD badly needs to catch Intel as far as performance/watt -- significantly improving fabric-level power consumption would bolster the company's competitive standing.
We certainly expect the company to talk up its consumer products, but Trinity's performance characteristics are well-mapped at this point.
is in an awkward position. It doesn't directly compete against Intel, but it's the face of an overwhelmingly successful mobile architecture that's increasingly threatened by Intel's Atom. With Intel talking up Clover Trail tablets and Atom going out-of-order at 22nm, ARM is going to focus on the Cortex-A15, its near-complete ownership of the smartphone/tablet market, and the successes of partners like Qualcomm, TI, Apple, and Nvidia.
ARM may or may not spend much bandwidth discussing its next-generation 64-bit architecture, ARMv8. The company's focus will be on painting Intel as a would-be competitor saddled with an x86 architecture that's too hamstrung by legacy issues to compete effectively. It's worth noting in advance that this simply isn't true. Intel's goal for smartphones in 2012 was to launch a platform that could compete with other midrange Android phones. It did so. Even if Intel's 22nm Atom slips to later in 2013, as is currently rumored, the company can afford the delay.
ARM's tremendous expertise and engineering chops shouldn't be discounted, but neither should Intel's long-term track record. As process node transitions become more difficult, Intel's ability to keep to a tick/tock model is an increasingly greater advantage.
isn't directly competing against Intel, but the company's iPhone 5 and possible new iPad event this week will steal some spotlight from Intel's own show. In this case, we don't expect much commentary one way or the other. Apple and Intel are important partners, but they're also pursuing disparate ends; each company wants to position its products at the center of consumers' digital lifestyles, and neither is eager to share the limelight.
In some ways, Apple's success underscores Intel's limited reach in the smartphone and tablet business; the iPhone drives far more business than the Mac or Macbook Pro. On the other hand, one could take Intel's current position at the heart of Apple's entire desktop line as evidence of how much times can change. Long-term, Intel wants to build a business for itself in handheld devices -- and it's got the manufacturing moxie to make that happen.