A new parallelism management scheme for multiprocessor systems

Abstract

Current multimedia and signal processing applications become more complex. This paper proposes a new parallelism management scheme that can explicitly deal with complex and general parallelism patterns. The parallelism description is based on a task flow graph representation interlaced with control commands. A graph management algorithm is proposed to extract eligible tasks and implement synchronization operations. We show that this management outperforms classical parallelism extraction in case of complex applications. Moreover, the parallelism description can be directly inserted in sequential programs without deep code modifications.