GPU Accelerated Scientific Computing: Evaluation of the NVIDIA Fermi Architecture; Elementary Kernels and Linear Solvers
Hartwig Anzt, Tobias Hahn, Vincent Heuveline, Björn Rocker · Repository KITopen (Karlsruhe Institute of Technology) · 2010
This paper evaluates the latest GPU generation from NVIDIA, called Fermi, with respect to the previous generation. The experiments include benchmarks of elementary kernels as well as of linear solvers applied to problems arising in the area of computational fluid dynamics. Both the consumer version of the hardware (GeForce GTX 480 and GTX 280) as well as the professional line (Tesla C2050 and C1060) are taken into account.