Load Balancing And Data Locality Via Fractiling: An Experimental Study

In order to fully exploit the power of a parallel computer, an application must bedistributed onto processors so that, as much as possible, each has an equal-sized,independent portion of the work. There is a tension between balancing processor loadsand maximizing locality, as the dynamic re-assignment of work necessitates access toremote data. Fractiling is a dynamic scheduling scheme that simultaneously balancesprocessor loads and maintains locality by exploiting the self-similarity...