Heterogeneous Computing Architectures Challenges and Vision

Heterogeneous Computing Architectures: Challenges and Vision provides an updated vision of the state-of-the-art of heterogeneous computing systems, covering all the aspects related to their design: from the architecture and programming models to hardware/software integration and orchestration to real-time and security requirements. The transitions from multicore processors, GPU computing, and Cloud computing are not separate trends, but aspects of a single trend-mainstream; computers from desktop to smartphones are being permanently transformed into heterogeneous supercomputer clusters. The reader will get an organic perspective of modern heterogeneous systems and their future evolution.

Preface Acknowledgements About the Editors Contributors 1 Heterogeneous Data Center Architectures: Software&Hardware Integration and Orchestration Aspects 1.1 Introduction 1.2 Backgrounds 1.3 Heterogeneous devices in modern data centers 1.4 Orchestration in heterogeneous environments 1.5 Simulations 1.6 Conclusions 2 Modular Operating Systems for Large Scale, Distributed and Heterogeneous Environments 2.1 Introduction 2.2 Modern Applications 2.3 Related Work 2.4 Modular Operating Systems 2.5 Analysis 2.6 Conclusion 3 Programming and Architecture Models 3.1 Introduction 3.2 Heterogeneous Programming Models 3.3 Heterogeneous Programming Languages 3.4 Heterogeneous Device Selection 3.5 Emerging Programming Models and Architectures 3.6 Ongoing European Projects 3.7 Conclusions 4 Simplifying Parallel Programming and Execution for Distributed Heterogeneous Computing Platforms 4.1 Introduction 4.2 StarSs: a Task-based Approach 4.3 Related Work 4.4 Conclusion 5 Design-time Tooling to Guide Programming for Embedded Heterogeneous Hardware Platforms 5.1 Introduction 5.2 Placer - Input and Output Metamodels 5.3 Graphical Front-End 5.4 Placement algorithm 5.5 DS-Explorer 5.6 The AquaScan Case Study 5.7 Related Works 5.8 Conclusion and Future Work 6 Middleware, Infrastructure Management and Self-Reconfiguration for Heterogeneous Parallel Architecture Environments 6.1 Introduction 6.2 Architecture 6.3 Adaptation Framework 6.4 Experimental Design 6.5 Evaluation 6.6 Related Work 6.7 Conclusion 7 A Novel Framework for Utilising Multi-FPGAs in HPC Systems 7.1 Introduction 7.2 The Framework 7.3 UNILOGIC & Hardware Platform 7.4 Runtime System 7.5 Accelerator Generation 7.6 Partial Reconfiguration 7.7 Conclusions 8 A Quantitative Comparison for Image Recognition on Accelerated Heterogeneous Cloud Infrastructures 8.1 Introduction 8.2 Computational approaches on Caffe 8.3 Performance evaluation and quantitative comparison 8.4 Conclusions 9 Machine Learning on Low-Power Low-Cost Platforms: an Application Case Study 9.1 Introduction 9.2 Backgrounds 9.3 Running CNNs on a bare-metal accelerator: a case study 9.4 Solution design 9.5 Experimental evaluation 9.6 Conclusion 10 Security for Heterogeneous Systems 10.1 Heterogeneous systems today 10.2 Heterogeneous systems issues 10.3 PDR secure framework 10.4 Conclusion 11 Real-Time Heterogeneous Platforms 11.1 Real Time task models 11.2 Architecture 11.3 Software support and programming models for future RT systems 12 Future Challenges in Heterogeneity 12.1 Introduction 12.2 The heterogeneous computing continuum 12.3 Processing architectures evolution 12.4 Programming challenges 12.5 Orchestration in heterogeneous environments 12.6 Conclusion Bibliography Index

Olivier Terzo is Head of the Advanced Computing and Electromagnetics (ACE) Research Area at Leading Innovation & Knowledge for Society (LINKS) Foundation. Karim Djemame is a Professor at the School of Computing and is the co-founder of the Heterogeneous Hardware & Software Alliance (HH&S), an initiative undertaken by the Transparent heterogeneous hardware Architecture deployment for eNergy Gain in Operation (TANGO) project. Alberto Scionti is a Senior Researcher in the Advanced Computing and Electromagnetics (ACE) Research Area at Leading Innovation & Knowledge for Society (LINKS) Foundation. Clara Pezuela is the Head of IT Market at Research and Innovation Group in Atos. She is also the President of PLANETIC, the Spanish technology platform for the adoption and promotion of ICT in Spain.