Static communication scheduling for minimizing collisions in application specific parallel systems

In applications requiring very high throughput or which have real-time deadlines, the use of parallel processing techniques has become widespread. Although there exists potential for vast performance gains, the communication overhead inherent in such systems can significantly lessen these gains. In this paper, with tightly-coupled architectures as the platform, the static communication scheduling of messages in the network is addressed. The compile time determination of when nodes should send their messages to other nodes is what is termed static communication scheduling. In parallel systems the static scheduling of computational tasks has been studied for some time, however, an in depth analysis of our problem is very new. This paper builds a framework based on the newly developed Collision Graph model. Using this model, the determination of an optimal schedule is proven to be NP-Complete. Several efficient algorithms are designed to deal with a general case model of message traffic, and experiments show a significant improvement over baseline approaches.