N-body-problem

COMP90025 - Parallel and Multicore Computing - 2020S2 - Assignment3

To markers

• report: docs/COMP90025_Assignment_03.pdf
• proposed hybrid naive O(n^2) approach: n2_hybrid/n2_hybrid.cpp
• proposed hybrid Barnes-Hut O(n logn) algorithm: nlogn_hybrid/nlogn_hybrid.cpp

directories

Content	Description
docs/	specification & report
random_body.cpp	genrate random 3D body
body_*.data	generated random body
generate_slurm.py	generate slurm scripts to be run on Spartan
draw.py	draw analysis diagrams
*.png	drawed figure in the report
n2_hybrid/	proposed hybrid O(n^2) algorithm
n2_hybrid_profile/	proposed hybrid O(n^2) algorithm's runtime profile
n2_openmpi/	parallel O(n^2) by pure OpenMPI
n2_openmpi_profile/	parallel O(n^2) by pure OpenMPI runtime profile
n2_sequential/	O(n^2) sequential
nlogn_hybrid/	proposed hybrid O(n logn) algorithm
nlogn_hybrid_profile/	proposed hybrid O(n logn) algorithm's runtime profile
nlogn_openmpi/	parallel O(n logn) by pure OpenMPI
nlogn_openmpi_profile/	parallel O(n logn) by pure OpenMPI runtime profile
nlogn_sequential/	O(n logn) sequential

How to run

• All files have instruction of how to run at the end of each file.

1. generate random bodies
2. python generate_slurm.py
3. cd n2_*/ or nlogn_*/ and sbatch *.slurm slurm scripts
4. python draw.py

submit

• sbatch xxx.slurm

spartan

• squeue -u xuliny
• scancel -u xuliny
• spartan-weather
• showq -a snowy
• squeue --partition snowy

Marks

• 16 out of 20 (80%)

• criteria	marks	comments
  Problem depth	20/20
  Writing style	15/20	Some poor wording here and there.
  Analysis	20/30	While what you did was good, it would have been better to study in more detail the reasons for certain behaviour and make stronger hypotheses.
  Results	15/20	Generally good but too many graphs
  Literature synthesis	10/10

• 

feedback

• 
• 
•