Contemporary multimedia applications usually consist of compute intensive components which process and communicate large data streams. These applications are commonly mapped on multiprocessor architectures which can meet the high performance requirements by exploiting the parallelism between the computations of the components. To allow such a mapping, the application should be written in a parallel form according to a chosen parallel model of computation (MoC). For streaming applications, the Process Network (PN) MoC represents a suitable choice. This book presents Compaan, an implementation of a novel four-step compilation approach that allows automatic translation of a sequential specification to a functionally equivalent PN specification for the affine nested loop programs.

Данное издание не является оригинальным. Книга печатается по технологии принт-он-деманд после получения заказа.