Single Node On-Line Simulation of MPI Applications with SMPI

Abstract

Simulation is a popular approach for predicting the performance of MPI applications for platforms that are not at one’s disposal. It is also a way to teach the principles of parallel programming and high-performance computing to students without access to a parallel computer. In this work we present SMPI, a simulator for MPI applications that uses on-line simulation, i.e., the application is executed but part of the execution takes place within a simulation component. SMPI simulations account for network contention in a fast and scalable manner. SMPI also implements an original and validated piece-wise linear model for data transfer times between cluster nodes. Finally SMPI simulations of large-scale applications on large-scale platforms can be executed on a single node thanks to techniques to reduce the simulation’s compute time and memory footprint. These contributions are validated via a large set of experiments in which SMPI is compared to popular MPI implementations with a view to assess its accuracy, scalability, and speed

Toward a Performance/Resilience Tool for Hardware/Software Co-design of High-Performance Computing Systems

“…es/computer-sciences/performance-tools/paraver) and Vampir [28] (http://www.vampir.eu). Also, SimGrid [29] and OMNeT++ [30] are multi-purpose simulation toolkits that have been used to investigate the runtime capabilities of future system architectures.…”

Scalable Multi-purpose Network Representation for Large Scale Distributed System Simulation

“…A clever instantiation leads to a very good approximation for large enough messages [10]. For small messages, very good results can be obtained with a piecewise linear model [11].…”

“…As explained in [8], the LogGPS model "ignores congestion in the network and assumes full effective bisection bandwidth", which is a valid hypothesis for several high-end computing platforms but not for cheaper Ethernet clusters or even torus-based networks as used in BlueGene or Cray machines. The model used in SIMGRID is an extension of the one proposed in [11] and accounts for such phenomenon as well as for network contention occurring in the core of the network. The importance of accounting for such phenomenon was demonstrated in [11] for MPI collective operations involving large amounts of data.…”

“…The model used in SIMGRID is an extension of the one proposed in [11] and accounts for such phenomenon as well as for network contention occurring in the core of the network. The importance of accounting for such phenomenon was demonstrated in [11] for MPI collective operations involving large amounts of data. LOGGOPSIM represents MPI executions as GOAL traces, i.e., DAGs of computations and communications.…”

Toward Better Simulation of MPI Applications on Ethernet/TCP Networks

“…Our model is implemented within SMPI [12], an opensource MPI implementation that connects real-world applications to the SimGrid [10] simulation framework. With SMPI, standard MPI applications written in C or FORTRAN can be compiled and run in a simulated network environment, and traces documenting computation and communication events can be captured without incurring errors from tracing overheads or improper synchronization of clocks as in physical experiments.…”