ABSTRACT
This chapter introduces compiler algorithms and a parallel computer system. It presents algorithms to compute existing programs as fast as possible, and also the possible speedup. The chapter outlines of the Harray system, which implements the algorithms. The Harray system being developed by the author’s group is a parallel processing system that implements the aforementioned theories. The controlled dataflow mechanism has been proposed to overcome the shortcomings of the data flow mechanism. Even though the data flow mechanism is suitable to expose the parallelism in a program naturally, it tends to consume a large amount of resources, such as the size of a memory unit for packet storage. The global control unit is the central controller that controls the execution order of blocks assigned to processors according to the global control graph. The synchroniztion memory unit synchronizes the completion of execution in the PEs.
