a b s t r a c tWe propose a numerical approach based on the Lattice-Boltzmann (LBM) and Immersed Boundary (IB) methods to tackle the problem of the interaction of solids with an incompressible fluid flow, and its implementation on heterogeneous platforms based on data-parallel accelerators such as NVIDIA GPUs and the Intel Xeon Phi. We explain in detail the parallelization of these methods and describe a number of optimizations, mainly focusing on improving memory management and reducing the cost of hostaccelerator communication. As previous research has consistently shown, pure LBM simulations are able to achieve good performance results on heterogeneous systems thanks to the high parallel efficiency of this method. Unfortunately, when coupling LBM and IB methods, the overheads of IB degrade the overall performance. As an alternative, we have explored different hybrid implementations that effectively hide such overheads and allow us to exploit both the multi-core and the hardware accelerator in a cooperative way, with excellent performance results.

No abstract

In this paper we propose and evaluate a set of new strategies for the solution of three dimensional separable elliptic problems on CPU-GPU platforms. The numerical solution of the system of linear equations arising when discretizing those operators often represents the most time consuming part of larger simulation codes tackling a variety of physical situations. Incompressible fluid flows, electromagnetic problems, heat transfer and solid mechanic simulations are just a few examples of application areas that require efficient solution strategies for this class of problems. GPU computing has emerged as an attractive alternative to conventional CPUs for many scientific applications. High speedups over CPU implementations have been reported and this trend is expected to continue in the future with improved programming support and tighter CPU-GPU integration. These speedups by no means imply that CPU performance is no longer critical. The conventional CPU-control-GPU-compute pattern used in many applications wastes much of CPU's computational power. Our proposed parallel implementation of a classical cyclic reduction algorithm to tackle the large linear systems arising from the discretized form of the elliptic problem at hand, schedules computing on both the GPU and the CPUs in a cooperative way. The experimental result demonstrates the effectiveness of this approach.

No abstract

No abstract