We are pleased to present you with this collection of material that captures a snapshot of the rich history of practice in Contemporary High Performance Computing. As evidenced in the chapters of this book, High Performance Computing continues to flourish, both in industry and research, both domestically and internationally. While much of the focus of HPC is on the hardware architectures, a significant ecosystem is responsible for this success.
Why I Edited This Book
My goal with this book has been to highlight significant systems and facilities in high performance computing. Early on, my main focus was proposed to be on the architectural design of important and successful HPC systems. However, I realized that HPC is about more than just hardware: it is an ecosystem that includes software, applications, facilities, educators, software developers, scientists, administrators, sponsors, and many other factors. This book is a snapshot of these contemporary HPC ecosystems, which are typically punctuated with a site’s flagship system. Broadly speaking, HPC is growing internationally, so I invited contributions from a broad base of organizations including the USA, China, Japan, Russia, Germany, and Switzerland.
My excitement about this book grew as I started inviting authors to contribute: everyone said “yes!” In fact, due to the limitations on hardback publishing, we had to limit the number of chapters that we could include in this edition.
As I explain in the introduction, the rate of change in HPC is accelerating. After 15 years of relative stability, scientists and facilities are looking to new architectures and programming models in order to achieve specific objectives in HPC while satisfying other constraints, like facility power and cooling.
Topic Selection and Organization
This book is organized into two parts. In Part I, the contributors provide perspective on contemporary HPC by examining significant trends in applications, performance, software, and hardware. Next, in Part II, the contributors provide snapshots of contemporary HPC ecosystems by site. While chapters in Part I do not follow a strict structure, I requested authors for chapters in Part II to include the following content:
1. Program background and motivation
2. Applications and workloads
3. Flagship system overview
4. Hardware architecture
5. System software
6. Programming system
7. Storage, visualization, and analytics
8. Data center/facility
9. System statistics
Some of the authors followed this outline precisely while others found creative ways to include this content in a different structure. Once you read the book, I think that you will agree with me that most of the chapters have exceeded these expectations, and have provided a detailed snapshot of their HPC ecosystem, science, and organization.
Helping Improve This Book
HPC and computing, in general, is a rapidly changing, large, diverse field. If you have comments, corrections, or questions, please send a note to me at [email protected].