GPU Computing Gems, Jade Edition, offers hands-on, proven techniques for general purpose GPU programming based on the successful application experiences of leading researchers and developers. One of few resources available that distills the best practices of the community of CUDA programmers, this second edition contains 100% new material of interest across industry, including finance, medicine, imaging, engineering, gaming, environmental science, and green computing. It covers new tools and frameworks for productive GPU computing application development and provides immediate benefit to researchers developing improved programming environments for GPUs.
Divided into five sections, this book explains how GPU execution is achieved with algorithm implementation techniques and approaches to data structure layout. More specifically, it considers three general requirements: high level of parallelism, coherent memory access by threads within warps, and coherent control flow within warps. Chapters explore topics such as accelerating database searches; how to leverage the Fermi GPU architecture to further accelerate prefix operations; and GPU implementation of hash tables. There are also discussions on the state of GPU computing in interactive physics and artificial intelligence; programming tools and techniques for GPU computing; and the edge and node parallelism approach for computing graph centrality metrics. In addition, the book proposes an alternative approach that balances computation regardless of node degree variance.
Software engineers, programmers, hardware engineers, and advanced students will find this book extremely usefull. For useful source codes discussed throughout the book, the editors invite readers to the following website:
Part 1: Parallel Algorithms and Data Structures – Paulius Micikevicius, NVIDIA 1 Large-Scale GPU Search 2 Edge v. Node Parallelism for Graph Centrality Metrics 3 Optimizing parallel prefix operations for the Fermi architecture 4 Building an Efficient Hash Table on the GPU 5 An Efficient CUDA Algorithm for the Maximum Network Flow Problem 6 On Improved Memory Access Patterns for Cellular Automata Using CUDA 7 Fast Minimum Spanning Tree Computation on Large Graphs 8 Fast in-place sorting with CUDA based on bitonic sort Part 2: Numerical Algorithms – Frank Jargstorff, NVIDIA 9 Interval Arithmetic in CUDA 10 Approximating the erfinv Function 11 A Hybrid Method for Solving Tridiagonal Systems on the GPU 12 LU Decomposition in CULA 13 GPU Accelerated Derivative-free Optimization Part 3: Engineering Simulation – Peng Wang, NVIDIA 14 Large-scale gas turbine simulations on GPU clusters 15 GPU acceleration of rarefied gas dynamic simulations 16 Assembly of Finite Element Methods on Graphics Processors 17 CUDA implementation of Vertex-Centered, Finite Volume CFD methods on Unstructured Grids with Flow Control Applications 18 Solving Wave Equations on Unstructured Geometries 19 Fast electromagnetic integral equation solvers on graphics processing units (GPUs) Part 4: Interactive Physics for Games and Engineering Simulation – Richard Tonge, NVIDIA 20 Solving Large Multi-Body Dynamics Problems on the GPU 21 Implicit FEM Solver in CUDA 22 Real-time Adaptive GPU multi-agent path planning Part 5: Computational Finance – Thomas Bradley, NVIDIA 23 High performance finite difference PDE solvers on GPUs for financial option pricing 24 Identifying and Mitigating Credit Risk using Large-scale Economic Capital Simulations 25 Financial Market Value-at-Risk Estimation using the Monte Carlo Method Part 6: Programming Tools and Techniques – Cliff Wooley, NVIDIA 26 Thrust: A Productivity-Oriented Library for CUDA 27 GPU Scripting and Code Generation with PyCUDA 28 Jacket: GPU Powered MATLAB Acceleration 29 Accelerating Development and Execution Speed with Just In Time GPU Code Generation 30 GPU Application Development, Debugging, and Performance Tuning with GPU Ocelot 31 Abstraction for AoS and SoA Layout in C++ 32 Processing Device Arrays with C++ Metaprogramming 33 GPU Metaprogramming: A Case Study in Biologically-Inspired Machine Vision 34 A Hybridization Methodology for High-Performance Linear Algebra Software for GPUs 35 Dynamic Load Balancing using Work-Stealing 36 Applying software-managed caching and CPU/GPU task scheduling for accelerating dynamic workloads
"Wen-mei Hwu: CTO of MulticoreWare, and is a professor at University of Illinois at Urbana-Champaign specializing in compiler design, computer architecture, computer microarchitecture, and parallel processing. He currently holds the Walter J. (""Jerry"") Sanders III-Advanced Micro Devices Endowed Chair in Electrical and Computer Engineering in the Coordinated Science Laboratory. He is a PI for the petascale Blue Waters system, is co-director of the Intel and Microsoft funded Universal Parallel Computing Research Center (UPCRC), and PI for the world's first NVIDIA CUDA Center of Excellence. At the Illinois Coordinated Science Lab, Dr. Hwu leads the IMPACT Research Group and is director of the OpenIMPACT project - which has delivered new compiler and computer architecture technologies to the computer industry since 1987. He previously edited GPU Computing Gems, a similar work focusing on NVIDIA CUDA."
Reviews for GPU Computing Gems Jade Edition
It wasn't until recently that parallel [GPU] computing made people realize that there are whole areas in computing science that we can tackle. ... When you can do something 10 or 100 times faster, something magical happens and you can do something completely different. -Jen-Hsun Huang, CEO, NVIDIA
