Load Balancing in the Macro Pipeline Multiprocessor System using Processing Elements Stealing Technique

The exponential increase in amount of processing power required by applications has been observed to be disproportionately leading the available processing power. Recent researches have shown that the key to balance this disproportionate is to shift the processor architecture towards multiprocessor or multicore architecture. In spite of the high performance of multiprocessor architecture, load imbalance is one of the key factors which mar its optimal performance. This paper focuses on how load balancing can be enforced on multiprocessor architecture in order to execute problems that are susceptible to domain decomposition. A technique for enforcing load balancing, called processing-elements-stealing, is proposed and a heuristic to implement this technique in multiprocessor architectures is developed. The objective of this paper is to show how processing element stealing technique can be used to prevent load imbalance in microprocessors architecture.