December 2017
Intermediate to advanced
718 pages
23h 14m
English
Given a specific algorithm or numerical method, there are several ways to express it in parallel computation. The choice of way will almost certainly be influenced by the hardware architecture present in the supercomputer, as well as how the community has addressed similar problems in the past. This chapter gives an overview of several parallel algorithms widely used in the scientific computing community, including fork–join, divide and conquer, halo exchange, permutation, embarrassingly parallel, manager–worker, and task dataflow.
Read now
Unlock full access