Improvements in computational power, whether on the PC, server or mobile device, have long rested with the processor. More powerful chips equaled more, better, faster data services.
But now that improvements to actual silicon have hit the brick wall save for the ability to pack more cores onto a single die, it’s no wonder that the industry is looking to other means to ramp up system capabilities.
One of those is the interconnect. As witnessed this week from Intel’s new silicon photonics system, there are a lot of gains to be had outside the chip. As the name suggests, the setup does away with copper-based electronic transfer of data between components in favor of photons over fiber optic links. The company says it can offer an initial boost from standard 10 Gbps throughput to 50 Gbps, with the potential to hit 400 Gbps in a short while. The technology also promises to extend the length of the interconnect, ushering in the possibility of silicon-to-silicon connectivity between servers, PCs, networking gear and other components.
Whenever you start to talk about fiber optics and lasers, however, you run into cost issues. Intel is aware of this, telling IDG that bringing the technology down to acceptable price-points is a crucial consideration as the technology moves from the lab to the field. The company is looking to produce a commercial product within the next decade, possibly a multichannel device that could see aggregate transfers speeds of more than 1 Tbps.
At the same time, all hope for the lowly electron is not lost. Researchers at Princeton University say they have discovered what they call a “super hero” electron capable of leaping over or breaking through the nanoscopic barriers on the surface of copper and other materials. The breakthrough could provide an entirely new avenue on which to transmit data as it offers the ability to control electrons on the surface of substances, such as antimony, as well as those in the interior.
Assuredly, there are countless other developments in laboratories around the world aimed at improving data transmission and handling, only some of which will arrive in actual products any time soon. But what is clear is that even though Moore’s Law may have finally reached its practical limit on the CPU, its corollaries will live on elsewhere in data technology.