The 280 pages of this book contain 17 papers drawn from the 2013 symposium on the same topic held at Nanyang Technological University, Singapore. The book is well edited and includes conclusions and future work discussions in most papers. The figures are well chosen, professionally created, and many are in color. Most papers emphasize empirical evidence of significant performance improvements of their subject algorithms using graphics processing units (GPUs).
GPUs were initially developed to accelerate the creation/update of graphical displays in video games. They work by dividing display processing among many different processors each working in parallel on only a part of the display. Today, these GPU systems on a chip can contain thousands of separate processors and many gigabytes of dedicated memory. Because they are inexpensive and present on most graphics cards, they are available for application to any algorithm where large blocks of data are processed in a parallel fashion. Because recasting domain-specific algorithms to execute well on GPUs is often difficult, such applications are the subject of much current research, including that reported in this book.
The 17 chapters in the book address several domains of interest, but focus on “big data.” Other areas covered by multiple chapters include engineering design, simulation, biomedical sciences, interactive computer graphics (deformation of constrained meshes, scientific visualization, animated virtual characters, virtual viewpoint rendering), and fast searching (approximate k-nearest neighbors and fast multi-keyword range).
Research papers on GPU applications are scattered among a wide variety of scholarly journals. It is refreshing to see a book like this that brings together research in multiple disciplines in a form that is accessible to so many readers. Read it because a few of the chapters pique your interest, but allow it to draw you into investigating and learning from adjacent fields.