CS 430/530: Parallel Computing

Fall 2014

CS 430/530: Parallel Computing (3-0-3)(F)(Even Years) Models of parallel computation. Fundamental design patterns used in parallel algorithms: embarrassingly parallel, partitioning, divide and conquer, software pipelining, synchronous computations and load balancing. Implementation of parallel programs using MPI, GPUs and Map-Reduce on parallel clusters. PREREQ: CS 253 and CS 321.

Mo We 12:00pm - 1:15pm

MEC 307

Prerequisite: CS 253 and CS 321

Notes	Handouts
Introduction (print version) Message Passing Model (print version) Introduction to MPI (print version) Design paradigms for parallel programs. Embarrassingly Parallel Computations (print version) (Exercises) Parallel Random Number Generation with the PRAND library (website, presentation) Partitioninig (print version) (Exercises) More on sending/receving in MPI (print version) Divide and Conquer (print version) Software pipelining (print version) Synchronized Computations (print version) Load Balancing (print version) Case Study I: Shortest Paths Case Study II: Pencil Beam Redefinition Algorithm Case Study III: Parallel Toolkit Library Open MP (Multi-threaded Parallelism) (print version) MapReduce Rationale (print version) Intro to MapReduce (with Hadoop) (print version) Further MapReduce Examples and Topics Last.fm case study (print version) Information Retrieval: Term Frequency-Inverse Document Frequency example (print version) Parallel Graph Algorithms for Social Networks (print version) Further MapReduce Design Patterns (print version) Data Warehouse and MapReduce (print version) Machine Learning and MapReduce (print version)	Lab Notes (PDF, HTML) Hadoop Port Assignments

Notes

Handouts