Light-speed Computer to Pack a Heavy Punch

From more advanced handheld devices to bigger, better number-crunchers to provide forecasts of climate change, the applications for faster computer processors are more numerous than […]

From more advanced handheld devices to bigger, better number-crunchers to provide forecasts of climate change, the applications for faster computer processors are more numerous than ever. In a world of finite resources and ever-growing energy demands, it’s increasingly important that these computers are made as efficient as possible. But the laws of thermodynamics limit the maximum number of transistors – the tiny switches at the heart of all computer processors – that can be crammed into small chips without overheating. This not only makes high-speed computers bulkier, it also prevents them from getting faster because of the long time required for data to travel between the processor and the computer memory.

But in new research published last month in Nature Photonics, an international team including researchers at Oxford University has found a way to bypass the communication bottleneck – using light. Beams of light carry information much faster than electronic circuits, but until now attempts to use them to create faster computers have been hampered by the need to convert information back into electronic form before it can be stored permanently in the memory. To get around this, the team has constructed new memory chips out of the same material used in rewritable CDs. The material changes phase in response to intense laser light, allowing users to store data directly from a light-based processor and simultaneously read from the memory using a lower-intensity beam.

Unlike previous solutions that required memory chips to be kept switched on for data to be preserved, the new method stores information indefinitely without needing a continuous power supply. The chip’s ultra-fast data storage and retrieval and relatively low power consumption could see it play a major role in the future of computing.

About Tobias Thornes