OpenACC Based GPU Parallelization of Plane Sweep Algorithm for Geometric Intersection

Abstract

Line segment intersection is one of the elementary operations in computational geometry. Complex problems in Geographic Information Systems (GIS) like finding map overlays or spatial joins using polygonal data require solving segment intersections. Plane sweep paradigm is used for finding geometric intersection in an efficient manner. However, it is difficult to parallelize due to its in-order processing of spatial events. We present a new fine-grained parallel algorithm for geometric intersection and its CPU and GPU implementation using OpenMP and OpenACC. To the best of our knowledge, this is the first work demonstrating an effective parallelization of plane sweep on GPUs.

We chose compiler directive based approach for implementation because of its simplicity to parallelize sequential code. Using Nvidia Tesla P100 GPU, our implementation achieves around 40X speedup for line segment intersection problem on 40K and 80K data sets compared to sequential CGAL library.

Appendix A Artifact Description Appendix

1.1 A.1 Description

Check-List (Artifact Meta Information)

• Algorithm:

  All algorithms are mentioned and described in the paper itself and can be referred to in Algorithms 1 and 3.

• Program:

  The Computational Geometry Algorithms Library (CGAL) and Geometry Engine Open Source (GEOS) were external libraries that were used.

• Compilation:

  Compilations were done using the g++ compiler and pgc++ compilers.

  for OpenACC: pgc++ -acc -ta=tesla:cc60 -o prog prog.cpp

  for OpenMP: g++ -fopenmp -o prog prog.cpp

  for CGAL: g++ -lcgal -o prog prog.cpp

  for GEOS: g++ -lgeos -o prog prog.cpp

• Hardware:

  Description of the machines used to run code can be found in Sect. 5.1 for further information.

• Publicly available:

  CGAL, GEOS, OpenMP, OpenACC, gcc and pgcc are all publicly available.

How Software Can Be Obtained (if Available). All of the software and code we used to build up our experiments were freely and publicly available. However, our code implementation can be found in the website: https://www.mscs.mu.edu/~satish/mpiaccgis.html.

Hardware Dependencies. To be able to get the most out of OpenMP, a multicore CPU would be needed. And to be able to run OpenACC kernels a GPU would be needed.

Software Dependencies. CGAL, GEOS, OpenMP and OpenACC libraries must be installed. Compilers like gcc and pgcc are also needed.

Datasets. Real world spatial data were used and datasets containing random lines were generated. Please refer to Sect. 5.1 for more information. Generated datasets are also posted in the website: https://www.mscs.mu.edu/~satish/mpiaccgis.html, however they can be generated on your own.

1.2 A.2 Installation

• 1. Configure the multicore CPUs and GPU to run on your system

• 2. Install the necessary libraries

• 3. Download or generate the necessary datasets

• 4. Download the code

• 5. Check that the datasets are in the proper directory pointed by the code, if not then fix it

• 6. Compile the code

• 7. Execute the compiled executable