Why Study Computing Architectures at Portland State University?
Device scaling, software complexity, adaptability, energy consumption, and fabrication economics indicate that the current computing paradigm cannot continue to deliver improvements at the rate we have enjoyed. To advance technologically and reap corresponding social and economic benefits, computing must become far more capable and energy-efficient. To enable new generations of computationally- and energy-efficient information processing engines over the next decade, our research focuses on a reconceptualization of the science and technology underlying the current approaches to co-design emerging devices, materials, interconnects, and physical phenomena.
The foundations of existing computing technology rely on equilibrium properties of closed thermodynamic systems where mass is conserved, and conventional computer science emphasizes trade-offs between memory resources and the number of time-steps needed to perform a given computation. However, the next generation of computing systems requires qualitative improvements over conventional CMOS, rather than merely continuing transistor scaling, especially to achieve thermodynamic efficiency. For example, newly available switching devices are being evaluated as building blocks for hybrid-CMOS and beyond-CMOS computing, and numerous devices exhibit exciting capabilities.
How will these new systems be modeled and programmed? Understanding and using new computing models based on unexploited phenomena will include developing hardware architectures, programming models, algorithms, runtime environments, and real-world applications. Our research focuses on developing novel device, circuit, and architecture concepts to exploit the unique physical properties of new materials. A long-term goal is to develop a generalized theory comprising digital, neuromorphic, and unconventional computing. Our proposed approach is exploratory, non-traditional, multi-disciplinary, and has the potential for breakthroughs and novel applications.
Our research has broad implications by enabling more computationally- and energy-efficient information processing systems for tomorrow's society, which continues to rely on mobile, battery-powered, edge devices as well as data centers.