|The Leading Source for Global News and Information Covering the Ecosystem of High Productivity Computing / March 30, 2007|
On April 22, 1977 General Telephone and Electronics sent the first live telephone traffic through fiber optics at the not-so-blazing speed of six megabits per second. Three decades later, optical communication is widespread. It's estimated that more than 80 percent of all the long-distance voice and data traffic is carried over optical fiber networks. Single fiber data rates are in the tens of gigabits per second.
But fiber bandwidth isn't everything. The bottleneck in optical communications comes when you have to convert the light into electrons. To do this today requires an array of discrete optical components. So while the cost of the optical fiber is cheap -- much cheaper than copper cable -- the cost of optical transceivers is not competitive with all-electronic solutions. The result is that at distances of 10 meters and under, optical communication is much less common than copper solutions. But as bandwidth requirements increase, the preference for optical interconnects over copper-based interconnects also increases, even at short distances.
Currently optical transceivers are built using a combination of expensive technologies. This includes discrete laser and photodetector components built from gallium arsenide or indium phosphide. These are connected to a printed circuit board along with dozens of other components and are interconnected by wires. The number of components that must be manufactured, assembled and tested makes current optical transceivers prohibitively expensive, except for long-haul applications.
Two recent announcements point the way to much less expensive optical solutions based on CMOS technology. Both IBM Research and Luxtera have demonstrated optical transceivers that take advantage of standard semiconductor technology to bring the core of optical communications onto silicon.
IBM researchers have built a prototype of an optical chip package which offers 160 Gbps of bandwidth -- 16 channels providing 10 Gbps each. This is eight times greater bandwidth than today's optical devices. The IBM optical transceiver has the driver and receiver circuits integrated onto a CMOS die. Other optical components, constructed from indium phosphide (InP) and gallium arsenide (GaAs), are added separately to the package, which is 3.25 by 5.25 millimeters in size. The optical transceiver is bump bonded in a flip chip assembly to attach the separate laser and photodiodes. The laser ends up sitting on top of the chip. The whole chipset uses just 2.5 watts of power.
The idea is to get the optics as close to the microprocessor as possible so that chip-to-chip communication can take advantage of the power savings and increased bandwidth offered by optical media.
Although IBM's press release mentions that the technology is capable of downloading a movie in a second, its first use will probably be in the data center. According to Marc Taubenblatt, senior manager for Optical Communications Group at IBM Research, this technology is especially interesting for HPC solutions, where the technology could be used to connect the microprocessor to I/O devices or the cluster interconnect fabric.
This will require printed circuit board manufacturers to add polymer waveguides to their boards and to develop compatible interfaces with IBM's optical circuitry. Mass manufacturing of optical printed circuit boards that incorporates the technology should enable low-cost optical solutions for node-to-node interconnects in a cluster.
"It's something that I think will take a few years for that ecosystem to become available in the market, but it's something we're actively working on," said Taubenblatt.
IBM presented the technology this week at the Optical Fiber Communication Conference & Exposition and the National Fiber Optic Engineers Conference, in Anaheim. The paper was titled "160 Gbit/s, 16-channel full-duplex, single-chip CMOS optical transceiver." What they described at OFC is the first generation of the technology. IBM has already prototyped the next generation in the lab that runs even faster. In this version each channel attains 12.5 Gbps reliably and can be pushed as high as 15 Gbps.
While IBM's technology may not show up in commercial products until 2010 or so, the Luxtera solution seems to be much further along. Luxtera, a startup company based in Carlsbad California, has been developing CMOS photonics since the company's inception in 2001.
Like IBM, Luxtera has produced an integrated optical chip on CMOS, demonstrating a four-channel 10 Gbps transceiver. On March 14, they announced success in integrating photodetectors onto the CMOS die, something not even IBM has achieved. The photodetector technology uses pure germanium around the optical waveguides for long-wavelength photodetection. With this feat, Luxtera will be able to produce an optical link with basically two components: the CMOS chip and the laser (or lasers depending on the number of light sources required). The tiny lasers are manufactured separately and mounted on the transceiver chip.
It is noteworthy that both Luxtera and IBM believe that silicon-based lasers, of the type Intel demonstrated in a prototype system in September of 2006, represents an immature technology. Even Intel projects that silicon-driven lasers will not be ready for commercial production until sometime in the next decade.
But all the remaining optical components can now be implemented on CMOS. The integration of the on-chip photodetectors has a number of advantages, according to Marek Tlalka, Luxtera's vice president of marketing. He says that by replacing discrete photodetectors with the on-chip version, these components essentially become free when produced in high volumes (approximately $3 for the discrete version versus a penny for the integrated kind). Also, since the photonic and electronic elements are in such close proximity to each other, the photodetectors can be made to be very sensitive. This enables a greater level of efficiency in the way laser photonics are used to drive the chip.
"Our photodetectors are about four times better than the best photodetectors that you can buy on the market today," said Tlalka.
Luxtera's goal is to bring to market a complete optical link -- two transceivers and the fiber -- for the price of a copper interconnect. Today's optical links built with discrete components costs around $700 per 10 Gbps, versus $200 for the copper version. Copper is actually fine at 10 Gbps for up to 2 meters. But once you get beyond 10 Gbps or a 2 meter reach, you need more electronic componentry and more copper to maintain the communications. Copper interconnects tend to max out at around 50 meters or so. By contrast, a single optical fiber strand can carry data up to two kilometers.
The other big downside to copper is its manageability. As communication distances increase, thicker copper cabling is required, increasing its weight and bending radius. Also, power consumption becomes an issue as data rates rise and communication distances increase.
Luxtera is aiming to deliver their first optical transceiver product in the second half of 2007. The 4 by 8 millimeter chip will be housed in a standard optical transceiver form factor so that it can be plugged into existing communication hardware. The first version will not include the just-announced integrated photodetector technology, but will instead use discrete components. Their CMOS chip is being fabbed by Freescale Semiconductor using a 130nm SOI process.
Luxtera's initial target market will be high performance computing. Tlalka thinks that HPC is a good entry point for them, since the volumes involved will allow them to systematically build up their supply chain and supplier base. He says they are working with Sun (a former partner in the DARPA HPCS program) and a couple of other OEMs who are interested in incorporating Luxtera's optical technology into their systems. During the ramp-up period, the company will manufacture and ship their own optical products. But Tlalka hints they may follow the Qualcomm model of establishing the technology on their own products and then later evolving into a chip and IP licensing company.
"Down the road, we're not going to rule out licensing deals, because there is so much opportunity for this technology," said Tlalka.
Beyond HPC application, he sees CMOS photonics hitting it big in the consumer market. As visual displays grow in size and support more color depth along with faster refresh rates, the required bandwidth increases rapidly. Connecting displays and projectors to PCs, DVD players and video games will soon demand much higher data communication rates. For example, version 1.3 of the High-Definition Multimedia Interface (HDMI) standard already specifies a bandwidth of 10.2 Gbps. In many cases, the distance between a display and other devices will be too far to be implemented with a reasonably priced copper solution at those data rates.
By achieving high levels of CMOS integration, Luxtera is able to apply the manufacturing and testing economic model of electronics to optics. Currently only the lasers and photodetectors are assembled independently, and presumably with the next generation, the integrated photodetectors will be included on-chip as well. The ability to streamline testing and assembly is the main reason Luxtera believes it can drive costs down into the copper interconnect range.
The other obvious advantage of using semiconductor technology is that it opens the door to integrating other electronic functionality onto the optical chip. For example, an Ethernet MAC (Media Access Controller) could be incorporated on the same die as the transceiver. If the technology is eventually licensed, third-party chip designers could include even more exotic communication features.
If CMOS integration does for optical components what it did for transistors, Luxtera could be in an enviable position. The demand for high-performance optical communications already exists in the HPC sector and is about to become much larger in the consumer market. By jumping on the Moore's Law curve for semiconductor technology, short-reach optical communication links may finally become widespread.