Random allocation of jobs with weights and precedence

Soumen Chakrabarti
1996 Theoretical Computer Science  
We analyze random allocation applied to irregular and dynamic task-parallel programs such as branch and bound. The precedence between jobs is revealed on-line, and the processing times of jobs are diverse and unknown before job completion. The objective is to assign jobs to processors and to schedule them to minimize makespan. We show that random allocation achieves makespan close to a natural lower bound. Some empirical experience with irregular parallel applications is reported. time at most
more » ... -l/P times optimal, where P is the number of processors. This, however, leads to a severe communication bottleneck at the processor where the pool of jobs resides. Our goal, therefore, is to study decentralized allocation which avoids such bottlenecks. The bottleneck can be relieved in a variety of ways, each of which reduces communication cost by sacrificing global load information and thus risking some load imbalance. We study work sharing, where busy processors forward jobs to random *
doi:10.1016/0304-3975(96)00036-9 fatcat:5fv5hxmmhbasxgj24p7l2ai3oq