Centers of Excellence for Computational Science and Engineering (COECSE)
Abstract
(U) The Centers of Excellence for Computational Science and Engineering (COECSE) will address the most difficult and fundamental challenges facing computing today. Computing has reached three walls of security, energy (power consumption) and programmability that cannot be overcome by traditional, evolutionary techniques. Security and energy-efficiency are difficult roadblocks for all current architectural approaches. Revolutionary new architectures, ranging from microprocessors, memory and interfaces to full-scale systems, are needed if we are to sustain the rate of advancement to which we have become accustomed. Languages that make programming current and future multi-core processors far more tractable for the average application developer are needed if we are to reap the benefits of emerging processor paradigms such as massive multi-core. The current approach to security, which attempts to retrofit security onto an evolving, imprecisely known, and increasingly complex (even non-deterministic) COTS infrastructure, is ad hoc and ineffectual – more systematic approaches are required. (U) Traditionally, computing has sought to overcome these three walls separately, with security, processing architectures, and programming languages developed in isolation and applied independently. The Centers of Excellence Program for Computational Science and Engineering will create research centers engaging academics and industry to explore and develop a more holistic approach to breaking down these walls. Examples of the types of research of interest include co-design approaches for hardware and software; parallel abstractions and new methods for expressing parallelism; software development environments for rapidly creating energy efficient embedded systems; computing components that have security “baked in” from the start for use at key points in the hardware and software stacks; provably secure clean-slate execution models; novel architectures for logic, memory, and data access to support secure execution; formal automated proof tools for security throughout the execution model; self-aware and learning capabilities to manage security at run-time; coordinated development of resiliency techniques (including detection and correction, fail-in-place self-healing, and learning); and new safe/secure computer languages and compilers.
Document Details
- Document Type
- Accomplishment
- Publication Date
- Oct 01, 2011
- Source ID
- 1036cd460738b55f46fc36c027c7e924