Title :
Grouping array layouts to reduce communication and improve locality of parallel programs
Author :
Shih, Tien-Pao ; Davidson, Edward S.
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., Michigan Univ., Ann Arbor, MI, USA
Abstract :
A data layout method, array grouping, is proposed to improve communication efficiency and cache utilization of parallel programs containing indirect array references or nonunit stride indexing. Conditions on where to apply this technique are specified in a series of theorems. The technique is then applied to a real finite element application. The experimental results show that communication is reduced by 15%, and data subcache misses by 40% on 56 processors of the KSR1 parallel computer
Keywords :
cache storage; concurrency control; parallel programming; KSR1 parallel computer; array grouping; cache utilization; communication efficiency; data layout method; data subcache misses; finite element application; indirect array references; locality; nonunit stride indexing; parallel programs; Application software; Concurrent computing; Context modeling; Data structures; Degradation; Finite element methods; Indexing; Multiprocessing systems; Pollution; Scalability;
Conference_Titel :
Parallel and Distributed Systems, 1994. International Conference on
Conference_Location :
Hsinchu
Print_ISBN :
0-8186-6555-6
DOI :
10.1109/ICPADS.1994.590375
