The Big Problem for computer architecture for the foreseeable future is the increasing latency of memory, relative to CPU speeds. Many schemes have been devised to ameliorate this problem, but none attempts to exploit the unique characteristics of the object-oriented model of computation, as embodied in Java(TM). Indeed, existing approaches do not even recognize the particular difficulties introduced by this model such as garbage collection, finer grain of locality, object-based synchronization, and so on. This project is investigating architectural innovations which can enhance the performance, scalability, availability and reliability of Java systems in the context of Sun's forthcoming microprocessors.
