One of the major impediments to the development of parallel computing
has been the lack of an abstract machine model. Such a model allows
portability of programs across architectures. In the absence of such
a model, programs have to be tailored to each specific architecture.
This project aims at developing such abstract machine models. This
project also has significant implications for conventional
microprocessors
with deep memory hierarchy. The issues of building spatial and
temporal locality into applications are common to both parallel and
uni-processors with deep memory hierarchies.