Skip to content

Scalability study of parallel scan in CUDA and OpenMP

Notifications You must be signed in to change notification settings

daltonbohning/Parallel-Scan-Scalability

 
 

Repository files navigation

Parallel Scan in CUDA and OpenMP

A scalability study of parallel scan using CUDA and OpenMP.

Authors

Dalton Bohning
Jordan Kremer

Report

Quicklink to Report.pdf

How to Run

The OpenMP version requires OpenMP to be installed and for "openmp" to be on the GCC path. The Brent-Kung version requires an NVIDIA CUDA-capable GPU with at least 2GB memory.

  • make benchmarks
    • Runs benchmarks for all versions
  • make benchmarks VERSION=iterative
    • Runs benchmarks for the single-threaded iterative version
  • make benchmarks VERSION=openmp_release
    • Runs benchmakrs for the OpenMP version
  • make benchmarks VERSION=brent_release
    • Runs benchmarks for the Brent-Kung version

The output can be found in a local file benchmarks.csv

CUDA (Brent-Kung)

  • Tested and works on wino.cs.pdx.edu on array sizes up to 268,435,456 (2^28) for section sizes 1024 and 2048.
    • The GPUs have ~2GB of memory, and an array of 2^28 float uses ~1GB. It's safer to not approach 2GB since the GPU is shared with the display device, etc.
  • The algorithm is done in-place on the device to conserve memory, and the output is copied to the output array, so a user would be oblivious to this.

OpenMP

Testing

make tests runs various tests on the CUDA and OMP versions and compares it to the iterative version.

Scalability Study

make benchmarks will produce a .csv with various array sizes, sections sizes of 1024 and 2048, and threads of 2, 4, 8, 16, and 32 for the CUDA, OMP, and iterative version.

About

Scalability study of parallel scan in CUDA and OpenMP

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Cuda 38.0%
  • Python 31.8%
  • C 21.6%
  • Makefile 8.6%