Green shoots of a sort: hardware support begins to sprout for Intel's Light Peak

Ensphere Solutions, a design-services shop that has recently entered the fabless semiconductor market with a range of multi-Gbit optical transceiver chips, today became one of the first vendors to announce hardware in support of Light Peak, Intel’s vision of a single optical interconnect cable for PCs and mobile devices. Ensphere is now sampling a 10 Gbit, two-channel transceiver die for use in the optical modules in a Light Peak connection.

In terms of core functionality the transceiver is fundamentally no different from other 10 Gbit transceivers, according to company vice president of marketing Al Gharakhanian. The die includes a VCSEL driver, a transimpedance amplifier for the PIN diode output, and a differential interface to the outside electronic world. While no 10 Gbit transceiver is a trivial design, Gharakhanian said, the interesting parts of this die are the diagnostics and power-management functions Intel envisions for their interface. The die is fabricated in 65nm TSMC CMOS.

Cost also had to be a key design consideration, Gharakhanian said. Intel envisions Light Peak linking two devices together for under $10, which is a long ways from current costs for a fast optical interconnect. The optical components themselves—the laser and PIN diode, and the fiber—already taking a great bite out of that budget, there’s not much to spend on electronics. The cost scenario didn’t get any rosier when the first optical cable vendor to announce a Light Peak product, Foci Fiber Optic Communications, said today that they had abandoned an attempt to use plastic optical fiber for performance reasons and had gone to treated glass fiber for the cable medium.

Just what the early hardware support for Light Peak means is unclear. There are several announced companies working with Intel to provide pieces of the total interconnect solution. But the architecture depends on a chip which Intel calls a router, which would reside on the host’s motherboard, pick up packet traffic from the computer and encode it in the appropriate serial protocol—PCIe, USB, HDMI, DisplayPort, or whatever, and ship it out to the optical module. Intel has not announced a schedule for the router chip, and without it the rest is pretty much just proof-of-concept prototyping. And then, as the current experience with USB 3.0 indicates, there is always the question of when Microsoft will notice that the bus needs a driver.

There is a further question. By pushing to make the PC platform dependent on low-cost optical interconnect, Intel may simply be attempting, as they suggest, to eliminate all but one of the cables in which today’s PC nests when it is in use. But there’s another possible explanation as well. Intel Labs has been working with some success on monolithic optical technology that would allow optical fibers to terminate right on the surface of a CMOS IC, with no need for separate optical components. Such technology applied to Light Peak would give Intel an insuperable advantage in cost. So it may be that the microprocessor giant is using smaller companies already in the optical transport space to pioneer the application for them, while Intel develops the integrated optical/transceiver silicon for with Light Peak was originally envisioned. One never knows.