logo-polimi
Loading...
Manifesto
Struttura Corso di Studi
Cerca/Visualizza Manifesto
Dati Insegnamento
Scarica il Manifesto
Regolamento didattico
Consulta il Regolamento
Elenco docenti
Strutture didattiche
Dati quantitativi
Internazionalizzazione
Orario Personalizzato
Il tuo orario personalizzato è disabilitato
Abilita
Ricerche
Cerca Docenti
Cerca Insegnamenti
Cerca insegnamenti degli Ordinamenti precedenti al D.M.509
Erogati in lingua Inglese

Legenda
Semestre (Sem)
1Primo Semestre
2Secondo Semestre
AInsegnamento Annuale
Lingua d'erogazione
Insegnamento completamente offerto in lingua italiana
Insegnamento completamente offerto in lingua inglese
--Non definita
Didattica innovativa
I CFU riportati a fianco a questo simbolo indicano la parte dei CFU dell'insegnamento erogati con Didattica Innovativa.
Tali CFU riguardano:
  • Cotutela con mondo esterno
  • Blended Learning & Flipped Classroom
  • Massive Open Online Courses (MOOC)
  • Soft Skills
Dati Insegnamento
Contesto
Anno Accademico 2017/2018
Scuola Scuola di Ingegneria Industriale e dell'Informazione
Corso di Studi (Mag.)(ord. 270) - MI (481) Computer Science and Engineering - Ingegneria Informatica
Piano di Studio preventivamente approvato T2A - COMPUTER SCIENCE AND ENGINEERING
Anno di Corso 1

Scheda Insegnamento
Codice Identificativo 051899
Denominazione Insegnamento ADVANCED TOPICS ON HETEROGENEOUS SYSTEM ARCHITECTURES
Tipo Insegnamento Monodisciplinare
Crediti Formativi Universitari (CFU) 5.0
Semestre --
Programma sintetico The course introduces students to the concept of Heterogeneous System Architectures, an increasingly adopted architectural paradigm in the entire computing spectrum (from embedded and mobile systems to high performance computing) by discussing their overall architecture and main opportunities and challenges in resources' programmability and management. Nowadays, trends in technology, fabrication processes and computing architectures are leading to the design and development of computing systems constituted by a relevant number of independent, heterogeneous processing resources, such as possibly-asymmetric multiprocessor CPUs, GPUs and (programmable) hardware modules. Indeed, the aim is to exploit the specific peculiarities of the various types of resources to achieve high-performance while leveraging power/energy consumption or other system-level aspects. However, system heterogeneity implies a relevant complexity in the usage of such a system to run a given workload, in terms of programmability (how to code an application to run on different resources) and runtime resource management (how to decide at runtime which processing unit to use for the application execution and how to tune architectural/execution parameters). To identify an optimal solution, system architects need to take into account the efficiency of systems' processing units, as well as the current workload and the working conditions. This course introduces students to the concept of Heterogeneous System Architectures (HSAs). First hardware and architectural aspects will be discussed, by presenting the overall architecture of an HSA and of the contained processing resources; in particular, we will give an overview of the organization of modern CPUs, GPUs and hardware accelerators (in particular FPGAs), and of the communication and memory infrastructure by discussing relevant aspects, such as memory hierarchies, cache coherency and virtual memory issues. Then, we will cover the software stack by discussing its main opportunities and challenges in resources' programmability and management. We will present the OpenCL programming paradigm and then provide some hints on the holistic solution for resource programmability and runtime management proposed by the HSA Foundation. Finally, an overview industrial and academic solutions for runtime resource management will be introduced; we will present the overall resource management framework and will discuss some state-of-the-art adaptation policies aimed at addressing different kinds of trade-off between application performance requirements and system's ones, in terms of energy/power consumption, resource utilization, temperature or lifetime reliability.

Orario: aggiungi e rimuoviScaglioneDocenteLingua offertaProgramma dettagliato
Da (compreso)A (escluso)
--AZZZZMiele Antonio Rosario
manifesti v. 3.3.7 / 3.3.7
Area Servizi ICT
09/07/2020