Efficient Parallelism via Hierarchical Tiling.
Larry Carter, Jeanne Ferrante, Susan Flynn Hummel · PPSC · 1995
Achieving high performance for applications that run on parallel systems in critical, yet difficult to achieve. Hierarchical tiling is a framework which can make this goal more readily attainable. It focuses on very structured computations where the bulk of the execution time is spent in nested loops. Hierarchical tiling determines which operations are to be executed on which processor in which order. Furthermore it choreographs the program, that is, it directs the movement of data between processing nodes and up and down the memory hierarchy of the target computer.