DataFlow Supercomputing Essentials Algorithms, Applications and Implementations
This illuminating text/reference reviews the fundamentals of programming for effective DataFlow computing. The DataFlow paradigm enables considerable increases in speed and reductions in power consumption for supercomputing processes, yet the programming model requires a distinctly different approac...
Uloženo v:
| Hlavní autor: | |
|---|---|
| Médium: | E-kniha |
| Jazyk: | angličtina |
| Vydáno: |
Cham
Springer Nature
2017
Springer International Publishing AG Springer International Publishing Springer |
| Vydání: | 1 |
| Edice: | Computer Communications and Networks |
| Témata: | |
| ISBN: | 3319661256, 9783319661254, 9783319661247, 3319661248 |
| ISSN: | 1617-7975, 2197-8433 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
Obsah:
- 3.3 Algorithm Description -- 3.3.1 Data Preparation -- 3.3.2 Binary Search Algorithm -- 3.4 DataFlow Algorithm Realization -- 3.4.1 Kernel Code -- 3.4.2 C Code -- 3.4.3 Manager Code -- 3.5 Performance Analysis -- 3.5.1 Test Results -- 3.5.2 Longer Input Sequence Analysis -- 3.5.3 Bottleneck Analysis -- 3.6 Conclusion -- References -- Part III Implementations -- 4 Introductory Overview on Implementation Tools -- 4.1 Introduction to dataflow Programming -- 4.2 DataFlow Engines (DFEs) and MaxJ Programming -- 4.3 Porting a Program to dataflow System -- 4.3.1 Program Analysis -- 4.3.2 Evaluating Different Architectural Options -- 4.3.3 Implementing the Kernels -- 4.3.4 Integration of the Accelerated Kernels Back into the Original Program -- 4.3.5 Debugging and Optimizing the Implementation to Get Maximal Performance -- 4.4 Interfaces -- 4.4.1 SAPI Single-DFE Single-Action Interface -- 4.4.2 MAPI Multiple Action Interface -- 4.4.3 DAPI DFE Creation API -- 4.5 Tools Overview -- 4.5.1 Maxtop -- 4.5.2 MaxIDE -- 4.5.3 MaxPower -- 4.5.4 Monitoring Dashboard -- 4.5.5 Maxdebug -- 4.5.6 Sliccompile -- 4.5.7 MaxSkins (DFEs Speak Your Language) -- 4.5.8 MaxOrchestrator -- References -- 5 DataFlow Systems: From Their Origins to Future Applications in Data Analytics, Deep Learning, and the Internet of Things -- 5.1 Introduction -- 5.2 High-Performance Computing Technology -- 5.2.1 State of the Control-Flow Technology -- 5.2.2 Special-Purpose Systems -- 5.2.3 DataFlow Systems -- 5.2.4 DataFlow Languages -- 5.2.5 Systolic Architecture -- 5.2.6 Multiscale dataflow Computing Approach -- 5.3 Ideas for New Applications of dataflow Systems -- 5.3.1 Software Distributed dataflow Systems for Big Data Analytics -- 5.3.2 Integration of MDC System with Apache Spark -- 5.3.3 Deep Learning Applications -- 5.3.4 The Internet of Things Applications -- 5.4 Conclusion -- References
- Intro -- Preface -- Dataflow Programming -- Dataflow Compilation -- Programming Maxeler -- Application Issues -- Conclusion -- Contents -- About the Authors -- Part I Algorithms -- 1 Implementing Neural Networks by Using the DataFlow Paradigm -- 1.1 Introduction -- 1.2 Existing Solutions -- 1.3 Essence of the dataflow Paradigm -- 1.4 Artificial Neural Networks -- 1.5 Single Layer Perceptron -- 1.5.1 Algorithm -- 1.5.2 The dataflow Implementation -- 1.6 Multilayer Perceptron Algorithm -- 1.6.1 Backpropagation Algorithm -- 1.6.1.1 Feed-Forward Phase -- 1.6.1.2 Feedback Phase -- 1.6.2 The dataflow Implementation -- 1.7 Performance Evaluation -- 1.7.1 Single Layer Perceptron -- 1.7.2 Multilayer Perceptron -- 1.8 Conclusion -- References -- Part II Applications -- 2 Solving the Poisson Equation by Using DataFlow Technology -- 2.1 Introduction -- 2.2 Mathematical Background -- 2.2.1 Boundary Conditions -- 2.3 Overview of Existing Solutions -- 2.3.1 Fourier Techniques -- 2.3.2 Iterative Techniques -- 2.3.2.1 Jacobi Method -- 2.3.2.2 Gauss-Seidel Method -- 2.3.2.3 Successive Over-relaxation Method -- 2.3.3 Multigrid Techniques -- 2.4 Overview of the Goal Solution -- 2.4.1 Maxeler dataflow Computing Technology Concept -- 2.4.2 Conditions of Applicability -- 2.4.3 Implementation Details -- 2.4.3.1 Kernel -- 2.4.3.2 PoissonSolver -- 2.4.3.3 Manager -- 2.4.4 User Guide -- 2.5 Measurements, Results, and Performances -- 2.6 Practical Applicability -- 2.6.1 Astrophysics -- 2.6.2 Electrostatics -- 2.6.3 Biology and Related Disciplines -- 2.6.4 Team Behavior Analysis -- 2.6.5 Other Applications -- 2.7 Conclusion -- References -- 3 Binary Search in the DataFlow Paradigm -- 3.1 Introduction -- 3.2 System Realization Requirements -- 3.2.1 The Problem -- 3.2.2 About Maxeler Technology -- 3.2.3 Conditions and Assumptions -- 3.2.4 Existing Solutions
- Index