Ruetsch, G: CUDA Fortran for Scientists and Engineers Best Practices for Efficient CUDA Fortran Programming

"This book is written for the Fortran programmer who wants to do real work on GPUs, not just stunts or demonstrations. The book has many examples, and includes introductory material on GPU programming as well as advanced topics such as data optimization, instruction optimization and multiple GPU programming. Placing the performance measurement chapter before performance optimization is key, since measurement drives the tuning and optimization process. All Fortran programmers interested in GPU programming should read this book." --Michael Wolfe, PGI Compiler Engineer