The Air Force’s Office of Scientific Research (AFOSR) wants to design a better class of mini-drones – micro air vehicles (MAVs) in military parlance – and they know just the scientist to help them do it.
Wu Feng, associate professor of computer science in the College of Engineering at Virginia Tech, was tapped by the Air Force to head up a large-scale, multi-disciplinary project aimed at reducing the time it takes to simulate the aerodynamics of these tiny aircrafts.
The Air Force is providing Feng and his team with $3.5 million in funding over three years with an option for a two-year extension and another $2.5 million.
Feng maintains they can “achieve substantial speed-up over current simulations and provide significantly better utilization of the underlying and co-designed hardware-software of a supercomputer.”
Feng is no stranger to innovation. A foremost expert in energy-efficient supercomputing designs, he co-founded the Green500 list in 2007. In 2011, Feng developed and built HokieSpeed, an accelerator-based supercomputer and one of the most energy-efficient systems of its kind.
Prior to joining Virginia Tech, Feng spent seven years at Los Alamos National Laboratory, where he designed and built the energy-sipping Green Destiny. This 2001-era supercomputer contained 240 nodes, occupied just five square feet of space, and consumed a mere 3.2 kilowatts of power – equivalent to the wattage consumed by two hair dryers.
The Air Force is looking to accelerator-based supercomputing in order to carry out computational fluid dynamics simulations in less time, but unlocking that ability will require cutting-edge software and hardware expertise. To address these multiple objectives, Feng has put together an impressive team of computer scientists, engineers and mathematicians from Virginia Tech and North Carolina State University.
Feng characterizes the project as necessitating advances in multi- and many-core parallel computing that will essentially “transform supercomputing.”
As part of the official announcement, Feng acknowledges the limits of Moore’s Law, but proposes that there is still “realizable performance [that] remains untapped.” Accelerators will be essential to their strategy. Feng notes that “coupling hardware-software co-design with advances in algorithmic innovation offers the promise of multiplicative speed-ups.”
Currently these miniature flying robots, a class of unmanned aerial vehicles (UAVs), are about five-inches across, but insect-sized drones are expected soon. In addition to their obvious military applications, the tiny air-bots will be useful in a variety of search and rescue operations.