DARPA seeking collaborators for the future of analogue computing
Fri 20 Mar 2015
The Defense Advanced Research Projects Agency (DARPA) has put out a Request for Information (RFI) inviting potential research collaborators to submit new ideas, methodologies and proposed technologies for leveraging analogue – non-digital, non-binary – processing methods in new research systems.
Vincent Tang, program manager at the Defense Sciences Office at DARPA explains “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,”
The new RFI is called ‘Analog and Continuous-variable Co-processors for Efficient Scientific Simulation’ (ACCESS), and is predicated on the principle that the diminishing returns available from a Moore’s Law-driven approach mean that the kind of computing power needed to analyse and simulate data on a grand scale is more likely to lie in probability-based computing solutions.
This is a field in which DARPA is renewing its interest, rather than making new or tentative steps. Five years ago it worked with MIT-born Lyric Semiconductor to produce the GP5 (GP standing for ‘General Probability’) analogue processor under the apparently shelved DARPA Analog Logic program. The GP5 is an application-specific integrated circuit (ASIC), and is intended to provide output and calculations which work in conjunction with digital systems rather than replacing them.
DARPA consolidated its work with Lyric in 2012 with the Unconventional Processing of Signals for Intelligent Data Exploitation (UPSIDE) program, which too seems to be now replaced by ACCESS.
Even the now-historical GP5 presents a power-consumption-to-efficiency ratio that explains why analogue computing has a future as well as a past, with tenfold comparative CPU speed backing up a hundredfold improvement in energy efficiency to comparable, digital-based processing paradigms. The GP5 ASIC employs its own PBSL machine-level language, and though its dedication to probability-based inference appears to put it in the same potential result-sphere as new research into quantum computing, its technical approach has nothing in common with that field of research.
No particular commercialised market sector is more interested in high-compute research than those which are involved with rendering large data sets, such as the videogames and visual effects sectors, and the GP5 caused some excitement in the VG community five years ago, when it seemed that the logic bottleneck could be broken by an analogue approach to character-generation.
DARPA’s interest in analogue computing seems to have been revived or bolstered by new correlative technology and research. Of the new initiative, DARPA says ‘[analogue computers’] potential to excel at dynamical problems too challenging for today’s digital processors may today be bolstered by other recent breakthroughs, including advances in microelectromechanical systems, optical engineering, microfluidics, metamaterials and even approaches to using DNA as a computational platform. It is conceivable, Tang said, that novel computational substrates could exceed the performance of modern CPUs for certain specialized problems, if they can be scaled and integrated into modern computer architectures,’
