Table of Contents
Module 4: Parallel Regions
Module 4: Parallel Regions
4.A Loop Level Paradigm
Performance of Loop-Level Model
Reducing Overhead
Coarser Grain with work-sharing
Using NOWAIT
Orphaned Directives
Statements Between Loops
Statements Between Loops (contd.)
Coarse Grain Work-sharing
4.B. Domain Decomposition
Domain Decomposition (contd....)
SPMD Programming
Implementing SPMD
Implementing SPMD (contd.....)
Handling Global Variables
Global private to thread
Threadprivate
Clauses on Parallel Regions
4.C. Comparisons with Message Passing
Message passing vs. OpenMP
Mixing OpenMP and Message Passing
4.D. Performance Considerations
Performance Optimization (contd.)
Point to point Synchronization
Example
Summary
|