ACCELERATED COMPUTING WITH HIP
Yifan Sun, Trinayan Baruah, David R Kaeli · LCC MAKS Press eBooks · 2024
This book primarily targets programmers looking to leverage GPUs to enhance computing performance. The textbook shows algorithms that take advantage of the latest heterogeneous computing platforms. The book, structured into eleven chapters and three appendices, provides a compact but comprehensive guide to HIP programming. The first four chapters, having reviewed the basics of graphics processors and parallel programming for these devices, introduce HIP programming, profiling, and debugging. Chapters 5-6 cover HIP programming patterns and stream-based computing. Chapters 7-10 present ROCm libraries, examples of using Hipify cross-platform tools, and then covers programming for a multi-GPU system and data center computing. The final chapter introduces third-party tools. The appendices cover AMD GPU assembly code and machine learning support in the ROCm environment. A special appendix on the installation and setup environment ofROCm has been added. The book provides sufficient examples of programming in HIP and ROCm to help the reader quickly master programming in HIP. The publication will be helpful for students, postgraduates, and teachers who use high-performance computing resources in their research and educational work.