Cutting-edge agency goes old-school

What: The Defense Advanced Research Projects Agency issued a request for information to technology developers about "old fashioned" analog computer processing technologies, with an eye to harnessing them in a new, extremely powerful way.

Why: DARPA said that while supercomputers have gotten faster and more powerful over the last 50 years and are capable of extremely complex, accurate simulations, they may have limits in those applications. In recent years, said the agency, even the best computer architectures haven’t been able to keep up with demand for the kind of simulation processing power needed to handle exceedingly complex design optimization and related problems. 

“The standard computer cluster equipped with multiple central processing units (CPUs), each programmed to tackle a particular piece of a problem, is just not designed to solve the kinds of equations at the core of large-scale simulations, such as those describing complex fluid dynamics and plasmas,” said Vincent Tang, program manager in DARPA’s Defense Sciences Office in a March 19 DARPA statement.

According to DARPA, those critical equations, called partial differential equations, describe fundamental physical principles like motion, diffusion, and equilibrium. However they involve a dense set of parameters and interactions that can confound individual CPUs attempts to break them and solve them in discrete pieces. A new processor could revolutionize simulation.

That new processing capability, it said, could have its roots in analog computers that solve equations by manipulating continuously changing values instead of discrete measurements. The technology was used in the Manhattan Project that developed the atomic bomb, as well as in the Norden bombsights that flew in World War II-era U.S. bombers. Analog computing was shoved aside in the 1950s and 1960s as transistor-based digital computers emerged and showed themselves to be more efficient for most kinds of problems.

Analog computers, according to Tang, have the potential to excel at dynamical problems too challenging for modern digital processors. They could also be spurred by advances in micro-electromechanical systems, optical engineering, microfluidics, metamaterials and even approaches to using DNA as a computational platform, he said.