This repository contains my reference examples of OpenMP API specification for parallel programming. It is based on the amazing Introduction to OpenMP by Tim Mattson (Intel).
hello.c — demonstrates how parallelisation is useless without synchronisation.
pi.c — calculates Pi in a good old one-threaded serial way. Used just as a starting point for performance improvements.
pi_par.c — basic usage of parallel pragma, single construct and atomic sync primitive. Forks a few worker threads and aggregates the final result in a thread safe way.
pi_par_loop.c — basic usage of parallel for and reduction to parallelize the loop and aggregate
the result. Compared to parallel approach above it keeps the code structure close to its serial version.
sharing.c — default, firstprivate, lastprivate, linear, none, private and shared data-sharing attribute clauses examples from OpenMP documentation.
utilise_cores.c — example of how to properly force exactly one thread per core.
...TBA
Compile and run all the examples:
make test
Complile and run one specific example:
make pi_par.exe && ./pi_par.exe
The .exe extension is used just becaue it's easier to ignore it in .gitignore.
Supported platforms are macOS and Linux (can be modified to run in Windows but nobody cares). See Makefile for details.