COINFLIPS

As part of a new microelectronics codesign research program, the Department of Energy’s Office of Science recently awarded the project $6 million over the next three years to develop the idea. The project team consists of Sandia National Laboratories, Oak Ridge National Laboratory, New York University, the University of Texas at Austin and Temple University in  Philadelphia. The team proposes that a “probabilistic computer” could  not only create smarter maintenance schedules but also help scientists analyze subatomic shrapnel inside particle colliders, simulate nuclear  physics experiments and process images faster and more accurately than  is possible with conventional computers.  

Codesign involves multidisciplinary  collaboration that takes into account the interdependencies among  materials, physics, architectures and software. Researchers also will  look at ways to incorporate machine learning methods. A  large numbers of devices, such as magnetic tunnel junctions and tunnel diodes, can be operated in a stochastic regime and incorporated into a scalable neuromorphic architecture that can impact a number of probabilistic computing applications, such as Monte Carlo simulations and Bayesian neural networks. 

 We have deterministic computing covered but we need probabilistic computing technologies:

• Modern microelectronics spends tremendous resources in enforcing determinism, while 
• Many applications desire probabilistic methods
• COINFLIPS is combining stochastic devices with neuromorphic architectures
• Co-design is proving invaluable in developing this novel paradigm for microelectronics.