SynchroSym

In a parallel computing scenario, the synchronization overhead, needed to coordinate the execution on the parallel computing nodes, can significantly impair the overall execution performance. Typically, synchronization is achieved by adopting a global synchronization involving all the nodes. In many application domains, though, a looser synchronization schema, namely, local synchronization, can be exploited, in which each node needs to synchronize only with a subset of the other nodes. SynchroSym is a simulator written in Matlab® purposely developed to assess the execution performance of parallel computing nodes that require to synchronize with each other by adopting either local or global synchronization. The parallel computation is considered to be step-based, and synchronization is required at the end of each step. In SynchroSym, the computing nodes can be organized along a mono-dimensional structure having N nodes, in which each node needs to synchronize with a number of left and right neighbours, and along a two-dimensional lattice with Nr rows and Nc columns in which each node needs to synchronize with its Von Neumann neighborhood (i.e., its north, south, east and west neighbors) or with its Moore neighborhood (which includes also the four diagonal neighbors). A simulation run is configured by specifying the number of involved computational nodes, the organization of the nodes, the number of computing steps composing the whole computation, and the probability distribution function used to assess the computation time for each node at each computational step.

Download files · Mono-dimensional simulator · Two-dimensional simulator

Authors: Franco Cicirelli (cicirelli@icar.cnr.it) - Andrea Giordano (giordano@icar.cnr.it) - Carlo Mastroianni (mastroianni@icar.cnr.it)