saidganim / nbody Goto Github PK

# N-body simulation algorithm implementation

Current N-body simulation is used for Parallel Programming Practical course at Vrije Universiteit Amsterdam.

Sequential version is provided by teachers. Parallel version has to be implemented by a student. A student has to use MPI for naive parallel implementation. A student is free to improve parallel version with any other libraries and ISA (OpenMP, CUDA, SIMD ISA). Up to student. Here I provide basic naive implementation first. I will try to imporove naive implementation with some improvements in future:
TODO:
	- Try to implement Intel intrinsics(SSE2 or AVX). - HARD TO IMPLEMENT, DATA STRUCTURES HAS TO BE CHANGED
	- Using OpenMP for some loops could improve performance.
	- Moving some computations to GPU accelerator also could help( very bonus , only if I have time for it) :)

Sequential code could be run at any computer. PDF file with instruction is contained in repo. Parallel code should be run on cluster to achieve real performance imporvement.  

# MPI assignment for the Parallel Programming Practical (PPP)


Content:
- nbody-par-bonus	Contains different optimizations for naive parallel version.

- nbody-par		Contains parallel version of N-Body algorithm. Naive version only

- nbody			Contains sequential version of the N-body algorithm
				Create parallel version of the N-body algorithm there 
				and name it nbody-par.

- docs			Place there your report.

- bin			Contains nbody-sanity-check (comparison with expected output)
				Do not submit programs which do not go through sanity-checks successfully.