Data Centers and High Performance Computing
Anno Accademico 2016/2017
La pagina principale del corso è raggiungibile qui.
Course Objectives
The Data Centers and High Performance Computing course aims at inroducing how to organize and effectively use high-end systems. To this end, modern computing architectures are first discussed at a single-chip level, with a special focus on massively-parallel architectures. Modern paradigms for concurrent and distributed programming are explained, which allow to fully exploit these modern architectures. The typical organization of data centers which host generic applications (such as web and data management ones) is presented, along with concepts related to their design and management. Emphasis will be given to the relationship between this organization and the Cloud Computing paradigm. The course will address as well topics related to transparency towards parallelism and distribution. Some topics will be associated with seminars given by renowned international companies.
Exam Rules
The exam is divided in two parts: a written test (5 questions, 2.30 hours) which will give 3/5 of the final mark; a practical project (2/5 of the final mark). The written test must be passed before starting the starting to work on the project.
We will record who attends the seminars. In case less than 60% of the seminars are attended, an additional essay on one of the topics presented at the seminars will be requested.
Programma del corso
- Advanced Computing Architectures - (Software Examples)
- Parallel architectures
- Multi-core/multiprocessor systems
- Cache-coherent systems
- NUMA implications
- Non-cache-coherent systems
- Concurrent Programming
- Correctness conditions
- Progress conditions and taxonomy
- Speed-up performance models
- Concurrent data structures
- Introduction to MPI (Message Passing Interface)
- Transactional Memory
- High Performance Computing Applications
- Continuous Simulation
- Monte Carlo Simulation
- Discrete Event Simulation
- Parallel and Distributed Simulation
- Data Centers
- Traditional data center architectures
- Cloud computing oriented data centers
- Large-Scale IT Service Management
- ITIL (Information Technology Infrastructure Library)
- Service Strategy
- Service Design
- Service Transition, Service Operation
- Continual Service Improvement
- Disaster Recovery e Business Continuity
Testi Consigliati
The following is a list of books which cover most of the topics of the course. In any case, lecture notes are enough to prepare the exam.
- Peter Pacheco, “An Introduction to Parallel Programming”. ISBN: 978-0-12-374260-5
- Maurice Herlihy, Nir Shavit, “The Art of Multiprocessor Programming”. ISBN: 978-0-12-370591-6
- Rachid Guerraoui, Paolo Romano, “Transactional Memory. Foundations, Algorithms, Tools, and Applications”. ISBN: 978-3-319-14720-8
- Alessandro Pellegrini, “Parallelization of Discrete Event Simulation Models”. ISBN: 978-8-89-853359-6
- Daniel Sorin, Mark Hill, David Wood, “A Primer on Memory Consistency and Cache Coherence”. ISBN: 978-1-60-845564-5
- Intel® 64 and IA-32 Architectures Software Developer’s Manual: Chapter 15 Programming With Intel® Transactional Synchronization Extensions
- Intel® 64 and IA-32 Architectures Optimization Reference Manual: Chapter 12 Intel® Tsx Recommendations