The Formal Functional Programming (FFP) machine is a novel computer architecture that has the potential of combining high performance with programming ease and flexibility. It is a direct implementation of a reduction model of computation. Program decomposition, task allocation, synchronization and scheduling are efficiently automated in the FFP machine, and therefore are removed from the programmer's concern. The objective of this research is to demonstrate that such a functional programming machine can be implemented in practical logic by designing and constructing an advanced prototype. The prototype will be a small- grain highly parallel computer that is built primarily from many copies of two building blocks: L-cells and T-cells. System software will be developed and the completed prototype will be evaluated. To help accomplish this objective, the resources of the Microelectronic Systems Laboratory will be utilized. This is a collaborative effort with Charles Molnar of Washington University. The program director considers the FFP machine to be a truly novel computer. Professor Mago is a leading authority in functional programming, and consequently is highly qualified to design and evaluate this machine. Co-principal investigator Vernon Chi has an excellent record in hardware-construction projects. Collaborator Charles Molnar at Washington University is well known for his expertise in asynchronous logic; hence, his efforts on a parallel grant from NSF will be supportive to this research. The program director strongly recommends funding this proposal.
