PARALLEL PROCESSING (CE-401

)
WEEK-WISE COURSE OUTLINE

COURSE INSTRUCTORS:
TOPICS

ENGR. SAQIB HUSSAIN SIDDIQUI (ASSISTANT PROFESSOR – COURSE COORDINATOR) HUMA JAMSHED (LECTURER), JAHANZEB KHAN (LECTURER)

COVERAGE Introduction
Week 1 ➢ Course Introduction ➢ Basic definitions

REFERENCE

Chapter 1

➢ Applications of parallel systems – ASSIGNMENT 1

Trends in Parallel architectures
Week 2
➢ Technological phases of processor design ➢ Flynn's taxonomy – QUIZ 1

Memory organization of parallel systems
◦ DMM – decoupling, considerations and ahead Weeks 3, 4 ◦ SMM – architecture, programming, advantage ◦ SMP and DSM architectures ◦ UMA, NUMA, ccNUMA architectures Chapter 2

Reducing memory access time
Week 5
◦ Multi-threading with its types ◦ Cache memory and consistency problem

Thread-level parallelism
Weeks 6, 7
◦ SMT (simultaneous multi-threading)

◦ Multi-core processors – architecture, types, advantages

Week 8 ➢ Introduction

MID-TERM EXAM
Interconnection networks
➢ Design criteria ➢ Types – static and dynamic networks Chapter 2

tinyurl. 15 marks for Mid-term exams 60 marks final paper . 15 ◦ Cost ◦ Speed-up ◦ Efficiency ◦ Amdahl's law ➢ Scalability and Gustafson's law Week 16 Chapter 4 Revision - COURSE BOOKS: Text: PARALLEL PROGRAMMING – FOR MULTI-CORE AND CLUSTER SYSTEMS (2010 edition) by: Rauber and Runger Reference: 'Reference slides' available on course website http://www.com/hpcmds THEORY MARKS DISTRIBUTION: 05 marks for Assignment / Quiz (best of two).➢ Properties ➢ Static networks – hypercube network ➢ Network embeddings (ring/mesh → hypercube) ➢ Dynamic networks Weeks 9 – 13 ◦ Bus systems ◦ Crossbar switches ◦ Multi-stage switch networks – omega networks ➢ Designing and routing hypercube static network ➢ Designing and routing omega dynamic network Chapter 2 Performance analysis ➢ MIPS and MFLOPS ➢ Performance metrics Weeks 14.