EagerMap

• DOI: 10.1145/3309711
• Publication Date: 2019-03-08T13:16:43Z
• Authors: Eduardo H. M. Cruz; Matthias Diener; Laércio L. Pilla; Philippe O. A. Na...
• Keywords: [INFO.INFO-DC]Computer Science [cs]/Distributed; [INFO.INFO-DS]Computer Science [cs]/Data Structures and Algorithms [cs.DS]; [INFO.INFO-DS] Computer Science [cs]/Data Structures and Algorithms [cs.DS]; 02 engineering and technology; Parallel; 004; Task mapping; and Cluster Computing [cs.DC]; Grid computing; [INFO.INFO-DC] Computer Science [cs]/Distributed, Parallel, and Cluster Computing [cs.DC]; 0202 electrical engineering, electronic engineering, information engineering; [INFO.INFO-DC]Computer Science [cs]/Distributed, Parallel, and Cluster Computing [cs.DC]

Communication between tasks and load imbalance have been identified as a major challenge for the performance energy efficiency of parallel applications. A common way to improve communication is increase its locality, that is, reduce distances data transfers, prioritizing usage faster more efficient local interconnections over remote ones. Regarding imbalance, cores should execute similar amount work. An important problem be solved in this context how determine an optimized mapping cluster nodes increases overall locality balancing. In article, we propose EagerMap algorithm task mappings, which based on greedy heuristic match application patterns hardware hierarchies can also consider load. Compared previous algorithms, faster, scales better, supports types computer systems, while maintaining same or better quality determined mapping. therefore interesting choice variety modern architectures.