Search

International audience; Nous considérons une classe de schémas Lagrange-Projection sur maillage décalé pour résoudre les équations d’Euler 1D. Les schémas proposés, formulés en énergie interne, sont d’ordre élevé en espace comme en temps, utilisables quelle que soit l’équation d’état. La présence des termes non-conservatifs est gérée en ajoutant une discrétisation de l’équation de l’énergie cinétique ainsi qu'une procédure de synchronisation entre l’énergie cinétique et l'impulsion. Cette procédure assure de résoudre le système formulé en énergie totale et permet au niveau discret de préserver globalement l’énergie totale ainsi qu'une bonne capture des chocs. Une extension nD sur grille cartésienne décalée de type C avec splitting directionnel d’ordre élevé est proposée. Des résultats numériques sont présentés jusqu’à l’ordre 8.

We propose a stochastic approach for calibration of mixing zone lengths in shock tube experiments. The methodology relies on taking into account uncertain initial data propagated through the basic multifluid Euler equations. In this work, the initial interface position is supposed uncertain, modeled by a stochastic process. The size of the mixing zone is then defined as the support of the probability density function of the stochastic process. This time dependent probability density function is estimated with non-intrusive generalized Polynomial Chaos, its support being in this case cheaply evaluated. This methodology relies on the application of an ergodic principle (Wiener, 1938) and generalizes linear perturbations analysis. It is applied in this Note to the calibration of several experimental results.

We first propose a new class of high-order finite volume schemes for solving the 1-D ideal magnetohydrodynamics equations that is particularly well-suited for modern computer architectures. Applicable to arbitrary equations of state, these schemes, based on a Lagrange-remap approach, are high-order accurate in both space and time in the non-linear regime. A multidimensional extension on 2-D Cartesian grids using a high-order dimensional splitting technique is then proposed. Numerical results up to fourth-order on smooth and non-smooth test problems are also provided.

We first propose a new class of finite volume schemes for solving the 1D Euler equations. Applicable to arbitrary equations of state, these schemes are based on a Lagrange-remap approach and are high-order accurate in both space and time in the nonlinear regime. A multidimensional extension on nD Cartesian grids is then proposed, using a high-order dimensional splitting technique. Numerical results up to 6th-order are provided.

We present here an extension to any order of accuracy of the schemes proposed in Daru and Tenaud [J. Comput. Phys. 193 (2) (2004) 563–594] for the linear advection equation in 1D. Such schemes are then used for a high-order generalization of the Godunov method in the case of the wave equation and the locally linearized Euler equations. To cite this article: S. Del Pino, H. Jourdren, C. R. Acad. Sci. Paris, Ser. I 342 (2006).

With the advent of the multicore era, parallel programming is becoming ubiquitous. Multithreading is a common approach to benefit from these architectures. Hybrid M:N libraries like MultiProcessor Communication (MPC) or MARCEL reach high performance expressing fine-grain parallelism by mapping M user-level threads onto N kernel-level threads. However, such implementations skew the debuggers' ability to distinguish one thread from another, because only kernel threads can be handled. SUN MICROSYSTEMS' THREAD_DB API is an interface between the debugger and the thread library allowing the debugger to inquire for thread semantics details. In this paper we introduce the USER LEVEL DB (ULDB) library, an implementation of the THREAD_DB interface abstracting the common features of user-level thread libraries. ULDB gathers the generic algorithms required to debug threads and provide the thread library with a small and focused interface. We describe the usage of our library with widely-used debuggers (GDB, DBX) and the integration into a user-level thread library (GNUPTH) and two high-performance hybrid libraries (MPC, MARCEL).

International audience; Message-Passing Interface (MPI) has become a standard for parallel applications in high-performance computing. Within a shared address space, MPI implementations benefit from the global memory to speed-up intra-node communications while the underlying network pro- tocol is exploited to communicate between nodes. But, it requires the allocation of additional buffers leading to a memory-consumption over- head. This may become an issue on future clusters with reduced mem- ory amount per core. In this article, we propose an MPI implementation upon the MPC framework called MPC-MPI reducing the overall memory footprint. We obtained up to 47% of memory gain on benchmarks and a real-world application.

Over the last decade, Message Passing Interface (MPI) has become a very successful parallel programming environment for distributed memory architectures such as clusters. However, the architecture of cluster node is currently evolving from small symmetric shared memory multiprocessors towards massively multicore, Non-Uniform Memory Access (NUMA) hardware. Although regular MPI implementations are using numerous optimizations to realize zero copycache-oblivious data transfers within shared-memory nodes, they might prevent applications from achieving most of the hardware's performance simply because the scheduling of heavyweight processes is not flexible enough to dynamically fit the underlying hardware topology. This explains why several research efforts have investigated hybrid approaches mixing message passing between nodes and memory sharing inside nodes, such as MPI+OpenMP solutions [1,2]. However, these approaches require lots of programming efforts in order to adapt/rewrite existing MPI applications. In this paper, we present the MultiProcessor Communications environnement (MPC), which aims at providing programmers with an efficient runtime system for their existing MPI, POSIX Thread or hybrid MPI+Thread applications. The key idea is to use user-level threads instead of processes over multiprocessor cluster nodes to increase scheduling flexibility, to better control memory allocations and optimize scheduling of the communication flows with other nodes. Most existing MPI applications can run over MPC with no modification. We obtained substantial gains (up to 20%) by using MPC instead of a regular MPI runtime on several scientific applications.

Thanks to recent advances in virtualization technologies, it is now possible to benefit from the flexibility brought by virtual machines at little cost in terms of CPU performance. However on HPC clusters some overheads remain which prevent widespread usage of virtualization. In this article, we tackle the issue of inter-VM MPI communications when VMs are located on the same physical machine. To achieve this we introduce a virtual device which provides a simple message passing API to the guest OS. This interface can then be used to implement an efficient MPI library for virtual machines. The use of a virtual device makes our solution easily portable across multiple guest operating systems since it only requires a small driver to be written for this device. We present an implementation based on Linux, the KVM hypervisor and Qemu as its userspace device emulator. Our implementation achieves near native performance in terms of MPI latency and bandwidth.

Our purpose is to provide the explanation of the successive waves which can occur in the general case when a wave propagates through a material with a non‐convex equation of state. We show that decrease of the sound speed introduces isentropic compression waves (ICW) inside shock waves (SW) and rarefaction shocks (RS) inside isentropic release (IR) fans. After basic phenomena explanations, we illustrate them on the example of a virtual material which gathers all these cases. Hydrocode calculations based on this material show the difficulties of high‐order numerical schemes to reproduce the physical features.

The development at CEA/DAMof a new AMR multi-physics hydrocode platform led to convincing results on a wide range of applications, from interface instabilities to charge computations in detonics.