CAMBRIDGE, UK — A prototype optical processor with a scalable, lensless design can crunch data faster than existing computers, according to its developer, Optalysys Ltd.
With a processing speed equivalent to 320 gigaflops, or 320 billion floating-point operations per second, the device is about the size of a desktop computer and uses significantly less power than traditional processors.
"Until now, meaningful optical processing technology has been impractical for a number of reasons," said Optalysys CEO Dr. Nick New. "Advancements in liquid crystal technology now permit numerical data to be dynamically entered into an optical system at high speeds and resolutions."
A prototype optical processor, right, beside a standard laptop. Courtesy of Optalysys.
The prototype demonstrates optical derivative functions — mathematical building blocks commonly used in complex engineering simulations such as weather prediction and aerodynamic modelling. It also performs correlation pattern matching, a technique used in DNA analysis and financial modelling.
The first project to use the technology starts next month. In collaboration with The Genome Analysis Centre (TGAC), in Norwich, England, Optalysys plans to build a Genetic Search System, or GENESYS. The project has been granted £0.5 million in funding from Innovate U.K., a government agency.
Genomic research is crucial for addressing diseases like cancer, as well as identifying and preventing pathogens that can decimate stable crops that provide global food security, according to TGAC.
However, public DNA databases are doubling in size every 18 months or less, which is making gene sequence searches too expensive for researchers. Such large databases require access to large high-performance computing resources that consume vast amounts of energy for power and cooling.
TGAC’s existing high-performance computing and related cooling systems collectively consume up to 130 kW of electricity. Using the optical processing system will result in energy savings of more than 95 percent per year, Optalysys and TGAC said.
Meanwhile, the Optalysys processor is set to play a role in another project led by the European Centre for Medium-Range Weather Forecasting (ECMWF) and involving 10 other organizations.
The aim of ESCAPE (Energy-efficient Scalable Algorithms for Weather Prediction at Exascale) is to develop extreme-scale computing capabilities for European operational numerical weather prediction (NWP).
The European Commission has favorably evaluated the proposal for ESCAPE for €4 million in grant funding through the Horizon 2020 Research and Innovation program. The grant is subject to final confirmation by the commission expected at the end of June.
Optalysys Chairman James Duez said the company plans to launch products in 2017 that will enable existing computers to achieve an equivalent processing rate of 9 petaflops, comparable to the fifth-fastest computer in the world today. By 2020 the company hopes to achieve larger systems capable multiple exaflops, or 1018 floating-point operations per second, Duez said.
For more information, visit www.optalysys.com and www.tgac.ac.uk.