Solving Hyperbolic PDEs using Accelerator Architectures

GPU;Cell Processor;Hyperbolic PDEs;Hardware Optimization;Applied Mathematics

Rostrup, Scott

University of Waterloo

关键词: GPU;    Cell Processor;    Hyperbolic PDEs;    Hardware Optimization;    Applied Mathematics;   

Others  :  https://uwspace.uwaterloo.ca/bitstream/10012/4518/1/rostrup-thesis-2009.pdf

瑞士|英语

来源: UWSPACE Waterloo Institutional Repository

【 摘 要 】

Accelerator architectures are used to accelerate thesimulation of nonlinear hyperbolic PDEs. Three different architectures, a multicoreCPU using threading, IBM’s Cell Processor, and Nvidia’s Tesla GPUs are investigated.Speed-ups of between 40-75× relative to a single CPU core in single precision are obtained using the Cell processor and the GPU.The three implementations are extended to parallel computing clusters by making useof the Message Passing Interface (MPI). The resulting hybrid-parallel code is investigatedfor performance and scalability on both a GPU and Cell computing cluster.