Readings in GPU: “Understanding performance bottlenecks in numerical kernels on GPUs” by Vasily Volkov

Volkov’s article is made up of a set of presentation slides, 53 in number, and specifically deals with performance bottlenecks for LU factorization and matrix multiplication. To the extent that these computations share numerical operations in common with other compute intensive tasks, the advice given in the article provides a general overview of performance bottlenecks [...]