Related topics

Is Apple behind Intel's speedy optical link?

Jobs-Otellini interconnect

Apple

The high-speed Light Peak optical interconnect that Intel unveiled at last week's developer confab was developed as a result of a CEO-to-CEO interconnect between Apple's Steve Jobs and Intel's Paul Otellini.

That is, if Engadget's "extremely reliable source" is correct. According to that website's report, the idea for Light Peak originated at Apple back in 2007 and was developed by Intel at Cupertino's urging.

This, of course, wouldn't be a unique event. Other now-widespread technologies have been conceived at One Infinite Loop then gone on to become industry standards. Think FireWire and OpenCL, for example, and the now-open source Grand Central Dispatch multicore-code helper.

The Reg believes that Engadget's mole knows what he's talking about, considering that Light Peak's coming-out party was hosted on a lucite-encased collection of boards running Mac OS X 10.6, better known as Snow Leopard - check out this video:

Light Peak also fits in well with Steve Jobs's mantra of "simplify, simplify, simplify." One cable to rule them all, as it were.

An additional interesting tidbit that wasn't announced during Light Peak's IDF rollout was noted by Intel Senior Fellow and Communications Technology Lab director Kevin Kahn during a separate session. "One of the things that was running over that optical fiber," he said, "was PCI Express."

Light Peak itself is an impressive piece of technology - hot-pluggable, full-duplex (10Gbps each way), multi-protocol - but it's only the beginning of what Kahn referred to as a "decade-type technology."

Currently based on well-tested Vixel optical-interconnect technology, Light Peak is slated to move up to Intel's breakthrough silicon photonics technology. "Relatively near future we'd like to get that up into the somewhere between 25 and 40 [Gbps] range," said Kahn, adding that "When we look at where do we go next with [Light Peak], we think the silicon photonics-type technology is probably the way we get to the higher data rates."

Intel is already working on speeds well beyond that. As Kahn told us, "If you're talking about applying the technology to very high-end systems, then 100 and up is interesting - and we've been kind of looking at that in the labs and have some prototypes just about done of some 100-class systems using silicon photonics. They're not cost-effective today, but will they be cost effective in few years? Probably yes."

And when Kahn says cost-effective, he doesn't mean only for high-end systems for deep-pocket high-end customers. He's talking mass acceptance. "This stuff has to cost a couple of bucks. Literally," he said. "And we think that's very achievable."

So when in the not-too-distant future we're hot-plugging tens of gigabits of connectivity into our portable devices, we may have Steve Jobs to thank for getting the ball rolling. ®

Sponsored: Designing and building an open ITOA architecture