Summary Communication and computation overlapping techniques have been introduced in the five‐dimensional gyrokinetic codes GYSELA and GKV. In order to anticipate some of the exa‐scale requirements, these codes were ported to the modern accelerators, Xeon Phi KNL and Tesla P 100 GPU. On accelerators, a serial version of GYSELA on KNL and GKV on GPU are respectively 1.3× and 7.4× faster than those on a single Skylake processor (a single socket). For the scalability, we have measured GYSELA performance on Xeon Phi KNL from 16 to 512 KNLs (1024 to 32k cores) and GKV performance on Tesla P 100 GPU from 32 to 256 GPUs. In their parallel versions, transpose communication in semi‐Lagrangian solver in GYSELA or Convolution kernel in GKV turned out to be a main bottleneck. This indicates that in the exa‐scale, the network constraints would be critical. In order to mitigate the communication costs, the pipeline and task‐based overlapping techniques have been implemented in these codes. The GYSELA 2D advection solver has achieved a 33% to 92% speed up, and the GKV 2D convolution kernel has achieved a factor of 2 speed up with pipelining. The task‐based approach gives 11% to 82% performance gain in the derivative computation of the electrostatic potential in GYSELA. We have shown that the pipeline‐based approach is applicable with the presence of symmetry, while the task‐based approach can be applicable to more general situations.

The current generation of the Xeon Phi Knights Landing (KNL) processor provides a highly multi-threaded environment on which regular programming models such as MPIjopenMP can be used. Many factors impact the performance achieved by applications on these devices: one of the key points is the efficient exploitation of SIMD vector units, and one another is the memory access pattern. Works have been conducted to adapt a plasma turbulence application, namely Gysela, for this architecture. A set of different techniques have been used: standard vectorization techniques, auto-tuning of one computation kernel, switching to high-order scheme. As a result, KNL execution times have been reduced by up to a factor 3. This effort has also permitted to gain a speedup of 2x on Broadwell architecture and 3x on Skylake. Nice scalability curves up to a few thousands cores have been obtained on a strong scaling experiment. Incremental work meant a large payoff without resorting to using low-level intrinsics.

Gyrokinetic modeling is appropriate for describing plasma turbulence in the core of Tokamaks, and the gyroaverage operator is a cornerstone of this approach. In a gyrokinetic code the gyroaveraging scheme needs to be accurate enough, but also requires a low computational cost because it is often applied on the main unknown, namely the 5D guiding-center distribution function, as well as on several 3D fields. The current gyroaverage implementation used in the Gysela code has recently been improved, enhancing the precision of the operator thanks to Hermite interpolation. In the present paper, we describe a new parallelization scheme for the gyroaverage operator. It mainly avoids costly transpositions of the full 5D function using halo exchange instead. Though the computational cost remains the same, the communication one is much smaller. The overall algorithm is also improved by cleverly interleaving communications and computations, thus allowing for a reduction of communication costs and a more efficient thread parallelization. The execution time with this algorithm is up to twice as fast as the previous version. The benefit of an improved scheme providing the overlap of communications by computations is also shown, again improving execution times. The description of the algorithms is given, together with an analysis of the achieved performance.

This paper describes optimization for high-dimensional stencil computations on accelerators involving complex memory access patterns, which appear in five dimensional fusion plasma turbulence codes, GYSELA and GT5D. They include different types of memory access patterns, the indirect memory access in GYSELA with a Semi-Lagrangian scheme and the strided memory access in GT5D with a Finite-Difference scheme. We focus on the affinity of the memory access patterns to accelerators such as GPGPUs and Xeon Phi coprocessors. On both devices, the Array of Structure of Array (AoSoA) data layout is preferable for contiguous memory accesses. It is shown that the effective local cache usage by improving spatial and temporal data locality is critical on Xeon Phi. On GPGPU, the texture memory usage improves the performance of the indirect memory accesses in the Semi-Lagrangian scheme. The reuse of registers by taking account of the physical symmetry of the Finite-Difference scheme reduces the amount of memory accesses. Through these optimizations, we achieve acceleration of 3.9 (8.1) on Xeon Phi (GPGPU) for the Semi-Lagrangian scheme and of 1.4 (3.9) on Xeon Phi (GPGPU) for the Finite-Different scheme with respect to the fully optimized codes on Sandy Bridge.

Summary Gyrokinetic simulations lead to huge computational needs. Up to now, the Semi‐Lagrangian code GYSELA performed large simulations using up to 65k cores. To understand more accurately the nature of plasma turbulence, finer resolutions are necessary, which make GYSELA a good candidate to exploit the computational power of future extreme scale machines. Among the Exascale challenges, the less memory per core is one of the most critical issues. This paper deals with memory management in order to reduce the memory peak and presents a general method to understand the memory behavior of an application when dealing with very large meshes. This enables us to extrapolate the behavior of GYSELA for expected capabilities of extreme scale machines. Copyright © 2014 John Wiley & Sons, Ltd.

SUMMARY We designed a fast parallel simulator that solves the acoustic wave equation on a graphics processing unit (GPU) cluster. Solving the acoustic wave equation in an oil exploration industrial context aims at speeding up seismic modeling and reverse time migration (RTM). We considered a finite difference approach on a regular mesh, in both two‐dimensional and three‐dimensional cases. The acoustic wave equation is solved in a constant density or a variable density domain. All the computations were carried out in single precision (both in the CPU reference implementation and in the GPU implementation), because double precision was not required in our context. We used Compute Unified Device Architecture to take advantage of the GPU computational power. We studied different implementations and their impact on the application performance. The described application handles all the steps of seismic modeling and RTM and is used to solve real‐world problems in an industrial production context. We obtained a speedup of 16 for RTM and up to 43 for the modeling application over a sequential code running on general‐purpose CPUs. A CPU rack versus a GPU rack comparison was described and showed a 4.3 speedup. Copyright © 2011 John Wiley & Sons, Ltd.

Abstract We are interested in a host–parasite system, i.e. the sea bass–Diplectanum aequans system. A discrete mathematical model is used to describe the dynamics of both populations. Our goal is notably to validate the model in the context of aquaculture. A deterministic numerical simulator and, recently, a stochastic simulator were developed to study this biological system. Parallelization is required because the execution times are too long. The Monte Carlo algorithm of the stochastic simulator and its three levels of parallelism are described. Analysis and performances, up to 256 processors, of a hybrid MPI/OpenMP code are then presented for a cluster of symmetric multi‐processor (SMP) nodes. Qualitative results are given for the host–macroparasite system simulation. Copyright © 2003 John Wiley & Sons, Ltd.