Evaluation of loop grouping methods based on orthogonal projection spaces

This paper compares three similar loop-grouping methods. All methods are based on projecting the n-dimensional iteration space J/sup n/ onto a k-dimensional one, called the projected space, using (n-k) linear independent vectors. The dimension k is selected differently in each method giving various results. The projected space is divided into discrete groups of related iterations, which are assigned to different processors. Two of the methods preserve optimal time completion, by scheduling loop iterations according to the hyperplane method. The theoretical analysis of the experimental results indicates the appropriate method, for specific iteration spaces and target architectures.

[1]  Nectarios Koziris,et al.  Lower Time and Processor Bounds for Efficient Mapping of Uniform Dependence Algorithms into Systolic Arrays , 1997, Parallel Algorithms Appl..

[2]  Weijia Shang,et al.  Time Optimal Linear Schedules for Algorithms with Uniform Dependencies , 1991, IEEE Trans. Computers.

[3]  Dan I. Moldovan,et al.  Partitioning and Mapping Algorithms into Fixed Size Systolic Arrays , 1986, IEEE Transactions on Computers.

[4]  Yves Robert,et al.  Mapping Uniform Loop Nests Onto Distributed Memory Architectures , 1993, Parallel Comput..

[5]  Jang-Ping Sheu,et al.  Partitioning and Mapping Nested Loops on Multiprocessor Systems , 1991, IEEE Trans. Parallel Distributed Syst..

[6]  Nectarios Koziris,et al.  Mapping nested loops onto distributed memory multiprocessors , 1997, Proceedings 1997 International Conference on Parallel and Distributed Systems.

[7]  Yves Robert,et al.  Constructive Methods for Scheduling Uniform Loop Nests , 1994, IEEE Trans. Parallel Distributed Syst..

[8]  Jih-Kwon Peir,et al.  Minimum Distance: A Method for Partitioning Recurrences for Multiprocessors , 1989, IEEE Trans. Computers.

[9]  Stephen H. Friedberg,et al.  Linear Algebra , 2018, Computational Mathematics with SageMath.