软件分布式共享存储系统的性能优化PDF|Epub|txt|kindle电子书版本网盘下载

软件分布式共享存储系统的性能优化PDF格式电子书版下载

注意：本站所有压缩包均有解压码： 点击下载压缩包解压工具

Chapter 1 Introduction1

1.1 Basic Idea of Software DSM2

1.1 Basic idea of software DSM3

1.2 Illustration of simple software DSM system5

1.2 Memory Consistency Model7

1.3 Cache Coherence Protocol9

1.4 Application Programming Interface10

1.5 Memory Organization11

1.6 Implementation Method13

1.6.1 Implementation Levels13

1.6.2 Granularity of the System15

1.7 Some Representative software DSMs16

1.1 Some Representative Software DSM Systems17

1.8 Recent Progress on Software DSM and Open Questions20

1.8.1 Software DSM-oriented Application Research20

1.8.2 Fine-grain vs. Coarse-grain Software DSM Systems22

1.8.3 Hardware Support for Software DSM System24

I.8.4 More Relaxed Memory Consistency Model25

1.8.5 SMP-Based Hierarchical Software DSM System27

1.9 Summarv of Dissertation Contributions29

1.10 Organization of the Dissertation33

2.1 Cache Coherence Protocol35

Chapter 2 Lock-Based Cache Coherence Protocol35

2.1.1 Write-Invalidate vs. Write-Update36

2.1.2 Multiple Writer Protocol37

2.1.3 Delayed Propagation Protocol38

2.1 Write merging in eager release consistency39

2.2 Comparison of communication amount in eager and lazy RC40

2.2 Snoopy Protocols40

2.3 Directory-Based Protocols42

2.3.1 Full Bit Vector Directory43

2.3.2 Limited Pointer Directory43

2.3.3 Linked List Directory44

2.3.4 Probowner Directory45

2.4 Lock-Based Cache Coherence Protocol46

2.4.1 Design Consideration46

2.4.2 Supporting Scope Consistency48

2.4.3 The Basic Protocol50

2.4.4 Correctness of the Protoeol51

2.1 Some Notations52

2.3 State transition digram of the lock-based cache protocol54

2.4.5 Advantages and Disadvantages54

2.2 Message Costs of Shared Memory Operations55

2.3 Comparison of Different Coherence Protocols56

2.5 Summary57

3.1 Introduction58

Chapter 3 JIAJIA Software DSM System58

3.2 Memory Organization59

3.1 Memory organization of CC-NUMA60

3.2 Memory organization of COMA61

3.3 Memory organization of JIAJIA63

3.4 Memory Allocation Example65

3.4 Programming Interface66

3.3 Lock-Based Cache Coherence Protocol66

3.5 Implementation68

3.5 Flow chart of threads creating procedure jiacreat()69

3.5.1 Starting Multiple Processes69

3.5.2 Shared Memory Management70

3.6 Flow chart of memory allocation jia_alloc(size)72

3.5.3 Synchronization75

3.7 Examples of nested critical sections78

3.5.4 Communication82

3.8 Communication between two processors85

3.6 Performance Evaluation and Analysis88

3.5.5 Deadlock Free of Communication Scheme88

3.6.1 Applications89

3.6.2 Performance of JIAJIA and CVM91

3.1 Characteristics of Benchmarks and Execution Results92

3.2 Eight-Processor Execution Statistics93

3.6.3 Confidence-Interval Based Summarizing Technique96

3.9 (？-μ)/？follows a t(n-1)distribution98

3.6.4 Paired Confidence Interval Method99

3.6.5 Real World Application:Em3d102

3.4 Execution Time,Scaled Speedup(S8)for Problem Scale 120×60×208103

3.3 Execution Time,Fixed Speedup(Sf)and Memory Requirement for Different Scales103

3.6.6 Scalability of JIAJIA105

3.10 Speedups of 8 applications under 2,4,8,16 processors106

3.7 Summary107

Chapter 4 System Overhead Analysis and Reducing108

4.1 Introduction108

4 2 Analysis of Software DSM System Overhead109

4.1 (a)General prototype of software DSM system.(b)Basic communication framework of JIAJIA110

4.2 Time partition of SIGSEGV handler and synchronization operation112

4 3 Performance Measurement and Analysis114

4.3.1 Experiment Platform114

4.1 Description of Time Statistical Variables114

4.3.2 Overview of Applications115

4.2 Characteristics of Applications116

4.3.3 Analysis116

4.3 (a)Speedups of applications on 8 processors.(b)Time statistics of applications117

4.3 Breakdown of Execution Time of These Applications118

4.3.4 The CPU Efiect121

4.4 (a)Comparison of speedups of fast CPU and slow CPU.(b)Effects of CPU speed to system overhead123

4.4.1 Reducing False Sharing124

4.4 Reducing System Overhead124

4.4.2 Reducing Write Detection Overhead126

4.4.3 Tree Structured Propagation of Barrier Messages127

4.4.4 Performance Evaluation and Analysis128

4.4 Characteristics of the Benchmarks129

4.5 Eight-Way Parallel Execution Results131

4.5 Breakdown of execution time134

4.5 Summary140

5.1 Background142

Chapter 5 Affinity-Based Self Scheduling142

5.2 Related Work145

5.2.1 Static Scheduling(Static)146

5.2.2 Self Scheduling(SS)146

5.2.3 Block Self Scheduling(BSS)147

5.2.4 Guided Self Scheduling(GSS)147

5.2.5 Factoring Scheduling(FS)148

5.2.6 Trapezoid Self Scheduling(TSS)149

5.2.7 Affinity Scheduling(AFS)149

5.2.8 Safe Self Scheduling(SSS)150

5.1 Chunk Sizes of Different Scheduling Schemes151

5.2.9 Adaptive Affinity Scheduling(AAFS)151

5.1 Basic framework of application155

5.3 Design and Implementation of ABS155

5.3.1 Target System155

5.3.2 Affinity-Based Self Scheduling Algorithm155

5.2 The Number of Messages and Synchronization Operations Associated with Loop Allocation157

5.4 Analytic Evaluation158

5.3 Description of the Symbols159

5.5.2 Application Description163

5.5.1 Experiment Platform163

5.5 Experiment Platforn and Performance Evaluation163

5.5.3 Performance Evaluation and Analysis166

5.4 The Effects of Locality and Load Imbalance(unit:second)167

5.2 Execution time of different scheduling schemes in dedicated environment:(a)SOR.(b)JI.(c)TC.(d)MAT.(e)AC169

5.5 The Number of Synchronization Operations of Different Scheduling Algorithms in Dedicated Environment170

5.6 The Number of Getpages of Different Scheduling Algorithms in Dedicated Environment170

5.7 System Overhead of Different Scheduling Algorithms in Dedicated Environment(Ⅰ)(second)171

5.8 System Overhead of Different Scheduling Algorithms in Dedicated Environment(Ⅱ)172

5.3 Execution time of different scheduling schemes in metacomputing environment:(a)SOR.(b)JI.(c)TC.(d)MAT.(e)AC176

5.10 The Number of Getpages of Different Scheduling Algorithms in Metacomputing Environment177

5.9 The Number of Synchronization Operations of Different Sc-heduling Algorithms in Metacomputing Environment177

5.11 System Overhead of Different Scheduling Algorithms in Metacomputing Environment(Ⅰ)178

5.12 System Overhead of Different Scheduling Algorithms in Metacomputing Environment(Ⅱ)179

5.4 Execution time with different chunk size under ABS scheduling scheme in metacomputing environment:(a)SOR.(b)JI.(c)TC.(d)MAT.(e)AC184

5.14 The Number of Getpages with Different Chunk Sizes in Metacomputing Environment185

5.13 The Number of Synchronization operations with Different Chunk Sizes in Metacomputing Environment185

5.15 System Overhead of Different Chunk Sizes in Metacomputing Environment186

5.6 Summary188

6.1 Introduction189

Chapter 6 Dynamic Task Migration Scheme189

6.2 Rationale of Dynamic Task Migration191

6.1 Basic framework of dynamic task migration scheme192

6.3.1 Computation Migration193

6.3 Implementation193

6.1 Definition of the Symbols194

6.3.2 Data Migration195

6.4 Home Migration196

6.5.2 Applications199

6.5.1 Experiment Platform199

6.5 Experimental Results and Analysis199

6.5.3 Performance Evaluation and Analysis200

6.2 Performance comparison:(a)execution time.(b)system overhead201

6.2 System Overheads in Unbalanced Environment202

6.3 System Overheads in Unbalanced Environment with Task Migration202

6.6 Related Work203

6.7 Summary205

Chapter 7 Communication Optimization for Home-Based Software DSMs207

7.1 Introduction208

7.2 Key Issues of ULN209

7.2.1 Communication Model210

7.2.2 Data Transfer211

7.2.4 Address Translation212

7.2.3 Protection212

7.2.5 Message Pipelining213

7.2.6 Arrival Notification213

7.2.7 Reliability214

7.1 Comparison of different communication substrate:(a)unreliable.(b)reliable214

7.2.8 Multicast215

7.3 Communication Requirements of Software DSMs215

7.4 Design of JMCL218

7.2 Interface description of JMCL218

7.1 Descriptions of JMCL Applications Programming Interface219

7.4.1 JMCL API219

7.4.2 Message Flow of JMCL221

7.4 Message transfer flow in UDP/IP222

7.3 Message transfer flow in JMCL222

7.5 Current State and Future Work224

7.6 Conclusion225

Chapter 8 Conclusions and Future Directions226

8.1 Conclusions226

8.2 Future of Software DSM230

Bibliography232