Trace-diven Co-simulation of High-performance Computing Systems using OMNeT++

In the context of developing next-generation high-performance computing systems, there is often a need for an “end-to-end” simulation tool that can simulate the behaviour of a full application on a reasonably faithful model of the actual system. Considering the ever-increasing levels of parallelism, we take a communication-centric view of the system based on collecting application traces at the message-passing interface level. We present an integrated toolchain that enables the evaluation of the impact of all interconnection network aspects on the performance of parallel applications. The network simulator, based on OMNeT++, provides a socket-based co-simulation interface to the MPI task simulator, which replays traces obtained using an instrumentation package. Both simulators generate output that can be evaluated with a visualization tool. A set of additional tools is provided to translate generic topology files to OMNeT’s ned format, import route files at run time, perform routing optimizations, and generate particular topologies. We also present several examples of results obtained that provide insights that would not have been possible without this integrated environment.