ARCHITECTURES AND MODELS FOR NUMERICAL METHODS

ARCHITECTURES AND MODELS FOR NUMERICAL METHODS

_
iten
Last update 29/07/2020 16:00
Code
98217
ACADEMIC YEAR
2020/2021
CREDITS
5 credits during the 2nd year of 10728 ENGINEERING TECHNOLOGY FOR STRATEGY (AND SECURITY) (LM/DS) GENOVA
SCIENTIFIC DISCIPLINARY SECTOR
ING-INF/01
LANGUAGE
English
TEACHING LOCATION
GENOVA (ENGINEERING TECHNOLOGY FOR STRATEGY (AND SECURITY))
semester
1° Semester
Teaching materials

AIMS AND CONTENT

LEARNING OUTCOMES

The course tackles the design and implementation of numerical algorithms for high performance computers in order to let students have a practical experience of the subject. In this framework, advanced architectures, parallel numerical algorithms, and their application to scientific and engineering problems are considered.

AIMS AND LEARNING OUTCOMES

The course aims to show computational models and computer architectures to let students understand their issues and deal with the implementation of numerical algorithms onto high performance computers. A wide range of topics is addressed,  from advanced architectures, parallel numerical algorithms, and their application to scientific and engineering problems.

Teaching methods

Lectures integrated by tutorials

SYLLABUS/CONTENT

  1. Review of basic computer architecture concepts: examples, software layers, processors, basic system architecture, Von Neumann machines, interrupts, CISC and RISC machines, digital signal processors (DSP), memory, DMA
  2. Parallel Computing: parallel and distributed computers, multiprocessor architecture, shared-memory and message-passing architectures, interconnection networks, interprocessor communication, performance analysis
  3. Parallel thinking: parallel algorithm design, parallel programming models, parallel performance. Introduction to parallel programming with python.
  4. Numerical models and algorithms:
    • Numerical simulation techniques
    • Intrroducion to Machine Learning algorithms. Random Forest algorithm.
    • Dense linear systems (vector and matrix products, LU factorization, triangular linear systems
    • Differential equations (ordinary and partial differential equations, numerical differentiation)
  5. Specialized architectures: Systolic arrays; CORDIC; GPU

RECOMMENDED READING/BIBLIOGRAPHY

  • Kaminsky, A., BIG CPU, BID DATASolving the World's Toughest Problems with Parallel Computing, CreateSpace Independent Publishing Platform; 1 edition (30 July 2016)
  • Benmammar, B. (2017), Concurrent, Real-Time and Distributed Programming in Java. Newark: John Wiley & Sons, Incorporated.
  • Zaccone, G. (2019), Python Parallel Programming Cookbook, 2nd edition, Packt Publishing. Birmingham, UK.
  • Lecture notes/slides provided by the teacher

TEACHERS AND EXAM BOARD

Ricevimento: By appointment after direct contact with the teacher.

Exam Board

RICCARDO BERTA (President)

ALESSANDRO DE GLORIA

ERMANNO DI ZITTI (President Substitute)

LESSONS

Teaching methods

Lectures integrated by tutorials

ORARI

L'orario di tutti gli insegnamenti è consultabile su EasyAcademy.

EXAMS

Exam description

Oral examination.

Assessment methods

The oral examination will address the mapping of parallel algorithms into parallel architectures for supporting numerical methods. In particular, at the beginning of the exam, each student will have the possibility to present the implementation of an algorithm of his choice into a specific concurrent and distributed architecture, pointing out key points and issues.