Compact and flexible linear-array-based implementations of apipeline of multiprocessor modules (PMMLA) for high throughputapplications

摘要

High throughput is required in many tasks, especially real-timeapplications. A logical structure of a parallel computing system forsuch applications is a pipeline of multiprocessor modules to be referredto as PMM. In this paper, a linear array based realization of PMM,referred to as PMMLA (PMM based on Linear Array) is proposed. The maindesign objective is to achieve uncompromised performance by a compactand flexible hardware structure. This paper describes the organizationand operation of a PMMLA, analyzes its performance in detail, andcompares it to other possible implementations theoretically and viaemulation on nCUBE/2