Google recently unveiled a new AI supercomputer that is showing the potential of optical interconnects. This technology can move data between computers at light speed, which could be integral in driving forward artificial intelligence and high-performance computing. Every major chipmaker is in agreement that optical interconnects are necessary for zettascale computing while still being energy efficient.
This opinion was further solidified by Google’s demonstration of their new AI supercomputer using this technology as an example of its capabilities. The test run showed how fast and reliable these connections can be when transferring large amounts of data from one computer to another, making it an ideal solution for future applications, such as machine learning or big data processing tasks. With this successful trial, researchers now have proof that optical interconnects can provide the speeds needed for next-generation technologies without sacrificing power efficiency or reliability.
Overall, Google’s AI Supercomputer shows just how powerful and useful optical interconnects could be if implemented correctly into existing systems and networks around the world; providing faster speeds than ever before while also reducing energy consumption levels drastically compared to traditional copper wires currently used today.
Read more at HPCwire
