Massively Parallel Solving of 3D Simplified PN Equations on Graphics Processing Units

International audienceThis paper presents a successful parallel implementation on Graphics Processing Units (GPUs) for the Simplified PN (SPN) calculations in the 3D case. For a nuclear operator such as EDF, the time required to compute nuclear reactor core simulations is rather critical. The SPN method provides a convenient trade-off between accuracy and numerical complexity and is used in several industrial simulations. The reduction of the computation time required to solve the eigenvalue problem goes through the parallelization of the algorithm. To solve the problem on distributed memory machines such as PC clusters, Domain Decomposition Methods have been investigated. Complementary to this approach, present work aims at using emerging massively parallel processors such as the GPUs. Based on a fine grained parallelism, this solution offers the opportunity to achieve good performances for a very low cost. Indeed, GPUs provide a huge computational power and some specific optimizations allow to near the hardware limits. Our implementation solves 3D SPN problems on GPUs 30 times faster than their sequential CPU couterpart