mizunototori / pyomp-1

OpenMP-like constructs for Python

PyOMP Usage Documentation

Usage:

Import PyOMP module
from omp import Omp

From this point onwards, till the end, a block of code refers to a function with the code inside it. This function shall not take any arguments unless its explicitly specified.
Use the parallel directive to start a parallel block.
@Parallel({"noThreads":4}

To parallelize a for loop, use the for directive. The function on which this directive (or decorator) is applied takes a list as an arugment, and the for loop must iterate through this
@For(List)

There is also ssupport for a single block and task blocks within
@Single @Task

Sections are created in the same way. With an outer block decorated with "sections" and containing blocks of "section"s within
@Sections @Section

A sample file is made available. Check sample.py

Results

Benchmarks of a matrix-matrix show that the performance is twice as better. (Lower is better).