网上多核 pdf下载

出版社科学出版社京东自营官方旗舰店

出版年 2020-06

页数 390页

装帧精装

评分 8.5(豆瓣)

⏰ 限时特惠 00:00:00

活动结束后恢复原价

纸质书参考价 ¥23

电子版限时价 ¥5.99 省 18 元

选择版本

基础版

PDF文件

¥4

内容简介

本篇主要提供网上多核电子书的pdf版本下载，本电子书下载方式为百度网盘方式，点击以上按钮下单完成后即会通过邮件和网页的方式发货，有问题请联系邮箱ebook666@outlook.com

内容简介

　　Chip multiprocessors have been the mainstream in computer architecture. Such processors have more than one core on a single chip to obtain higher performance and lower power consumption. However， when more and more cores are integated onto the chip， the increasing number of cores makes the communication become the center of the on-chip architecture. It has become a new challenge on how to utilize these integrated cores in a single area with high efficiency. This book provides the explorative research on the above problems. There are four main topics discussed in this book. The first is the hybrid on-chip structure， which consists of the on-chip network and the on-chip bus. The bus is used to connect the local neighbours， and the network is used to connect the remote nodes. And then this book provides discussions on on-chip structure and optimizations including the dynamic reconfigurable network， the
　　design of critical path-driven routers and the transmission bypass optimization. They are used to achieve better performance. The third topic focuses on the on-chip memory design， which is called the on-chip networked memory system. The memories are connected by the network with distributed memory management units. At last， the mapping algorithms are discussed， which aim to map the tasks to the on-chip cores with high efficiency and low-power consumption. This book has provided some works innetwork-based manycore systems. The designs and algorithms provided in this book are potential
　　solutions for multicore/manycore architecture. It can be used as the reference for future work and for the researchers who focus on the computer architecture.

内页插图

Contents
Preface
1 Introduction to NoC 1
1.1 Development of Computer Architecture 1
1.2 Chip Multiprocessor 3
1.3 On-chip Structure and NoC 6
1.4 Summary 10
2 Hybrid Network and Bus On-Chip Interconnection 11
2.1 Introduction 11
2.2 Hybrid On-chip Interconnection with NoC and the Bus 13
2.2.1 Motivation 13
2.2.2 On-Chip Structure 14
2.2.3 Support for Thread Scheduling 15
2.2.4 Experiments and Results 16
2.3 Dynamic Configurable On-Qiip Network with the Hybrid Bus and Networks 18
2.3.1 Motivation 18
2.3.2 Bus/NoC Hybrid Interconnection 19
2.3.3 Component Design 21
2.3.4 Experiment and Results 23
2.4 Summary 28
3 On-Chip Structure and Optimizations 30
3.1 Introduction 30
3.2 Dynamic Reconfigurable Networks for I/O-Supported Parallel Applications 31
3.2.1 Background 31
3.2.2 Architecture Design 33
3.2.3 Implementation 37
3.2.4 Experiments and Analysis 42
3.3 Critical Path-Driven Routers for the On-Chip Network 46
3.3.1 Background 46
3.3.2 Motivation 48
3.3.3 Architecture 51
3.3.4 Implementation 53
3.3.5 Experiments and Analysis 56
3.4 Transmission Bypass Optimization for On-Chip Cores 60
3.4.1 Background 60
3.4.2 Motivation 61
3.4.3 Design 62
3.4.4 Implementation 65
3.4.5 Experiments and Analysis 68
3.5 Summary 75
4 On-Chip Networked Memory System for NoC 77
4.1 Introduction 77
4.2 Network Main Memory Architecture for NoC 78
4.2.1 Background 78
4.2.2 Motivation 80
4.2.3 Basic NMM Architecture 81
4.2.4 Management of NMM and Software Model 85
4.2.5 Experiments and Analysis 87
4.3 Distributed Memory Management Units Architecture for NoC 94
4.3.1 Background 94
4.3.2 Motivation 98
4.3.3 Architecture Model 100
4.3.4 Experiments and Analysis 104
4.4 Summary 108
5 Efficient Task Mapping Algorithm with Low-Power Design for NoC 110
5.1 Introduction 110
5.2 Efficient Task Mapping Algorithm with Power-Aware Optimization for NoC 111
5.2.1 Background Ill
5.2.2 Motivation 112
5.2.3 System Model 114
5.2.4 Proposed Algorithm Design 119
5.2.5 Experiments and Analysis 129
5.3 Energy-Efficient Design of the Microkernel-Based On-Chip OS for NoC 133
5.3.1 Background 133
5.3.2 Motivation 134
5.3.3 Design Overview 136
5.3.4 Distributed On-Chip Operating System 139
5.3.5 Experimental Results and Analysis 144
5.4 Summary 151
6 Conclusions 153
References 156

查看全部↓

前言/序言

　　Processors are the basis of the powerful computing systems. When single-core processors cannot be improved for higher performance and lower power consumption， chip multiprocessors or multicore processors are designed to solve this problem by the cooperation of the relatively simple cores or processing elements （PEs）. When the traditional bus cannot provide communication efficiency for many cores， the on-chip network is proposed as a solution to connect these cores as a network on chip （NoC）. NoC can provide efficient on chip communications among many cores. However， the new architecture provided by NoC means emerging challenges including the on-chip network structure， the on-chip components， the memory organization and the mapping algorithm， etc. Such factors have important impacts on the overall performance of NoC-based many core systems.
　　This book focuses on the above problems and proposes related explorative solutions. Hybrid on-chip network designs are described first. Both the bus and the network have their advantages and disadvantages. They are used by different methods to organize the on-chip network， in which the bus is used to provide high bandwidth and short communication time， and the network is used to provide long-distance communication. And then the description is provided on the optimizations of the on-clup components including on-chip routers and cores. The routers are designed to forward the messages on the critical paths first. And a bypass is provided to the special stage of the on-chip core's pipeline to reduce waiting time. Then， the network's main memory structure is proposed to provide the distributed memory access response to reduce the pressure of the high frequency of memory references. The distributed memory management unit （MMU） structure is also designed to provide memory management for higher efficiency. At last， the corresponding mapping algorithms are proposed， which can schedule the tasks to the cores effectively with low-power consumption. And the microkemel is also used to support the mapping algorithm.
　　This book will provide some work in network-based many core systems. We hope the designs and algorithms can be a reference to the researches in NoC. NoC itselfis in rapid changes， which always makes it necessary for us to pay close attention to the progress of the researches. It also needs the new ideas， the corresponding design， the implementation， the analysis and the optimization. In the future， NoC will be more powerful to provide higher performance.

查看全部↓