AIMS AND CONTENT

TEACHING METHODS

Class, laboratory, discussions, personal study

SYLLABUS/CONTENT

• Computer performance, benchmarks.
• Pipeline processors: internal organization, hazards and their impact on performance, hardware and software techniques for reducing hazards and/or their effects, instruction-level parallelism, dynamic instruction scheduling, branch prediction, speculative execution.
• Multiple-issue and VLIW processors (sketch).
• Cache memory and how to take advantage of it.
• MIMD multiprocessors: Amdahl law and the impact of communication latency.
• Shared-memory and shared-address-space MIMD: organization, scalability issues (memory contention, cache coherence), synchronization in shared address space. Memory consistency models. Highl-level programming with OpenMP.
• Message-passing MIMD: organization, support to communication and synchronization, the SPMD paradigm, the MPI standard, parallel programming examples with MPI.
• Load balancing techniques (sketch).

See web pages of the course for updates

RECOMMENDED READING/BIBLIOGRAPHY

John Hennessy, David Patterson: Computer Architecture: A Quantitative Approach, Morgan Kaufmann Publishers.

TEACHERS AND EXAM BOARD

Exam Board

GIUSEPPE CIACCIO (President)

GIOVANNI CHIOLA

FRANCESCO MASULLI

LESSONS

Class schedule

PARALLEL COMPUTING

EXAMS

EXAM DESCRIPTION

Oral examination and project discussion