Efficient PDE Solvers on Modern Hardware with Applications in Medical and Technical Sciences Export

@phdthesis{citeulike:5711842, abstract = {The dissertation presents a category theory based approach to the design and implementation of parallel numerical algorithms on modern hardware architectures. Furthermore an efficient parallel implementation of the preconditioned conjugate gradient algorithm with algebraic multigrid preconditioner (PCG-AMG) is presented. Specifically the parallel implementation of the algebraic multigrid (AMG) algorithm is analyzed and a simplified parallelization approach is derived. A main result of the dissertation is the implementation of the PCG-AMG algorithm on graphics processing units (GPUs) capable of outperforming parallel implementations of the algorithm on high performance cluster computers.}, author = {Liebmann, Manfred}, citeulike-article-id = {5711842}, keywords = {gpu, pde\_solvers}, month = {August}, posted-at = {2009-09-03 06:56:15}, priority = {2}, school = {University of Graz}, title = {Efficient PDE Solvers on Modern Hardware with Applications in Medical and Technical Sciences}, year = {2009} }

TY - THES ID - citeulike:5711842 L3 - citeulike-article-id:5711842 N2 - The dissertation presents a category theory based approach to the design and implementation of parallel numerical algorithms on modern hardware architectures. Furthermore an efficient parallel implementation of the preconditioned conjugate gradient algorithm with algebraic multigrid preconditioner (PCG-AMG) is presented. Specifically the parallel implementation of the algebraic multigrid (AMG) algorithm is analyzed and a simplified parallelization approach is derived. A main result of the dissertation is the implementation of the PCG-AMG algorithm on graphics processing units (GPUs) capable of outperforming parallel implementations of the algorithm on high performance cluster computers. TI - Efficient PDE Solvers on Modern Hardware with Applications in Medical and Technical Sciences PB - University of Graz KW - gpu KW - pde_solvers AU - Liebmann, Manfred PY - 2009/08// ER -