- GCC
- CUDA 9.2 or higher
- PGI 19.10 or higher
- Python3 (Numpy and Matplotlib), used for benchmark only.
This project provides multiple implementations of a N-bodies interraction simulation.
The implementations have been made in single-thread, OpenMP, OpenACC and CUDA:
- nbody.c: First implementation made by the professor.
- nbody.cu: Naive CUDA portage of nbody.c
- nbodyomp.c: OpenMP version of nbody.c
- nbodyacc.c: OpenACC version of nbody.c
- nbodysoa.c: Structure of Array version of nbody.c
- nbodysoa.cu: CUDA portage of nbodysoa.c
These verions can be compiled and run with the Makefile provided in nbody.
Please run cd nbody && make all to compile all the sources. You can also use make help to see all the available commands.