Название: Hardware Accelerators in Data Centers Автор: Christoforos Kachris, Babak Falsafi, Dimitrios Soudris Издательство: Springer ISBN: 3319927914 Год: 2018 (2019 Edition) Страниц: 280 Язык: английский Формат: pdf (true), epub Размер: 14.9 MB
This book provides readers with an overview of the architectures, programming frameworks, and hardware accelerators for typical cloud computing applications in data centers. The authors present the most recent and promising solutions, using hardware accelerators to provide high throughput, reduced latency and higher energy efficiency compared to current servers based on commodity processors. Readers will benefit from state-of-the-art information regarding application requirements in contemporary data centers, computational complexity of typical tasks in cloud computing, and a programming framework for the efficient utilization of the hardware accelerators.
Emerging cloud applications like machine learning, artificial intelligence (AI), deep neural networks, and big data analytics have created the need for more powerful data centers that can process huge amounts of data without consuming excessive amounts of power. To face these challenges, data center operators have to adopt novel architectures with specialized high-performance energy-efficient computing systems, such as hardware accelerators, that can process the increasing amount of data in a more energy-efficient way. To this end, a new era has emerged; the era of heterogeneous distributed computing systems that consist of contemporary general-purpose processors (CPUs), general-purpose graphic processing units (GP-GPUs), and field-programmable gate arrays (FPGAs or ACAP Adaptive Compute Acceleration Platform). The utilization of several heterogeneous architectures poses several challenges in the domain of efficient resource utilization, efficient scheduling, and resource management.
In this book, we have collected the most promising and the most recent research activities in this emerging domain of heterogeneous computing that is based on the efficient utilization of hardware accelerators (i.e., FPGAs). The book contains 13 system-level architectures that show how to efficiently utilize hardware accelerators in the data centers to face emerging cloud applications.
1 Introduction 1 2 Building the Infrastructure for Deploying FPGAs in the Cloud 9 3 dReDBox: A Disaggregated Architectural Perspective for Data Centers 35 4 The Green Computing Continuum: The OPERA Perspective 57 5 Energy-Efficient Acceleration of Spark Machine Learning Applications on FPGAs 87 6 M2DC—A Novel Heterogeneous Hyperscale Microserver Platform 109 7 Towards an Energy-Aware Framework for Application Development and Execution in Heterogeneous Parallel Architectures 129 8 Enabling Virtualized Programmable Logic Resources at the Edge and the Cloud 149 9 Energy-Efficient Servers and Cloud 163 10 Developing Low-Power Image Processing Applications with the TULIPP Reference Platform Instance 181 11 Energy-Efficient Heterogeneous Computing at exaSCALE—ECOSCALE 199 12 On Optimizing the Energy Consumption of Urban Data Centers 215 13 Improving the Energy Efficiency by Exceeding the Conservative Operating Limits 241 Index 273