Pierre-Nicolas Clauss

Single Node On-Line Simulation of MPI Applications with SMPI

Simulation is a popular approach for predicting the performance of MPI applications for platforms that are not at ones 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 simulations 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.

Iterative computations with ordered read-write locks

We introduce the framework of ordered read-write locks, ORWL, that are characterized by two main features: a strict FIFO policy for access, and the attribution of access to lock-handles instead of processes or threads. These two properties together allow applications to have a controlled pro-active access to resources and thereby to achieve a high degree of asynchronicity between different tasks of the same application. For the case of iterative computations with many parallel tasks which access their resources in a cyclic pattern, we provide a generic technique to implement them by means of ORWL. We show that the possible execution patterns for such a system correspond to a combinatorial lattice structure and that this lattice is finite if and only if the configuration contains a potential deadlock. In addition, we provide efficient algorithms: one that allows for a deadlock-free initialization of such a system and another one for the detection of deadlocks in an already initialized system.

Experimenting Iterative Computations with Ordered Read-Write Locks

This paper presents the first experimental results of the use of our new adaptive tool for synchronization, based on ordered read-write locks, ORWL. They provide a new synchronizing method for data-oriented parallel algorithms and are particularly suited for iterative pipelined algorithms with out-of-core data. We conducted experiments with the classic benchmarking Livermore Kernel 23 algorithm to validate the theoretical model and measure the efficiency of the first available implementation of ORWL in the PARXXL library. They show that this tool is able to efficiently control an IO bound application running on 64 parallel POSIX threads with tight data dependencies between them.