[Back]

M. Faraj, A. van der Grinten, H. Meyerhenke, J. Träff, C. Schulz:
"High-Quality Hierarchical Process Mapping";
Talk: 18th International Symposium on Experimental Algorithms, SEA 2020 - Online Conference, Catania, Italy; 2020-06-16 - 2020-06-18; in: "18th International Symposium on Experimental Algorithms, SEA 2020", S. Faro, D. Cantone (ed.); Schloss Dagstuhl - Leibniz-Zentrum für Informatik, LIPIcs 160 (2020), ISBN: 978-3-95977-148-1; 4:1 - 4:15.

Partitioning graphs into blocks of roughly equal size such that few edges run between blocks is a frequently needed operation when processing graphs on a parallel computer. When a topology of a distributed system is known, an important task is then to map the blocks of the partition onto the processors such that the overall communication cost is reduced. We present novel multilevel algorithms that integrate graph partitioning and process mapping. Important ingredients of our algorithm include fast label propagation, more localized local search, initial partitioning, as well as a compressed data structure to compute processor distances without storing a distance matrix. Moreover, our algorithms are able to exploit a given hierarchical structure of the distributed system under consideration. Experiments indicate that our algorithms speed up the overall mapping process and, due to the integrated multilevel approach, also find much better solutions in practice. For example, one configuration of our algorithm yields similar solution quality as the previous state-of-the-art in terms of mapping quality for large numbers of partitions while being a factor 9.3 faster. Compared to the currently fastest iterated multilevel mapping algorithm Scotch, we obtain 16% better solutions while investing slightly more running time.

Process Mapping, Graph Partitioning, Algorithm Engineering

http://dx.doi.org/10.4230/LIPIcs.SEA.2020.4

Project Head Jesper Larsson Träff:
Process Mapping