k-Means clustering algorithm

University Project of @AlexStingaci and @pgnsamu

Read the handout and use the sequential code as reference to study. Use the other source files to parallelize with the proper programming model.

Use the input files in the test_files directory for your first tests.

Table of Contents

• Installation
• Configuration
• Usage
• Testing
• License

Installation

Step-by-step instructions on how to get the development environment running.

# Example command
git clone https://github.com/pgnsamu/KMEANS.git
cd KMEANS

Configuration

Edit the first lines in the Makefile to set your preferred compilers and flags for both the sequential code and for each parallel programming model: OpenMP, MPI, and CUDA.

To see a description of the Makefile options execute:

make help 

Usage

There are 4 files that will be compiled whenever you use

make all

After that you will need to run the executable of KMEANS.c and set the output in output2D.inp using this method then you can check if the results are fine with the sequential program

./KMEANS_seq test_files/input100D2.inp 1000 3 50 0.01 output2D.inp

For running parallel programs

OMP

./KMEANS_omp test_files/input100D2.inp 1000 3 50 0.01 output2D2.inp

MPI

mpirun -np 4 ./KMEANS_mpi test_files/input100D2.inp 1000 3 50 0.01 output2D2.inp

For MACOS

export OMPI_CC=gcc-14   # For Open MPI
export MPICH_CC=gcc-14  # For MPICH

MPI+OMP

MAKEFILE

mpicc -fopenmp -o KMEANS_mpi_omp KMEANS_mpi_omp.c -lm

run

mpirun -np 4 ./KMEANS_mpi_omp test_files/input64D.inp 50 100 0.01 0.001 output2D2.inp

Unlimited Test

ulimit -s unlimited

If you wanna run the program with debug variables you can do

make debug 

Testing

We are planning to use github autotest, but for now before pushing you need to run tests by yourself.

The test scripts are confronto.py and extractTime.py. Before running confronto.py you need to modify num_esecuzioni default value that is 10

num_esecuzioni = 10

execute the sequential program (with the command above), then you have to run the program modified (still using the command above) and last you have to run confronto.py

python confronto.py

this will check if the result of the parallel program are the same of the sequential program and save time of each execution in tempi_esecuzione.txt

License

EduHPC 2023: Peachy assignment

(c) 2022 Diego García-Álvarez, Arturo Gonzalez-Escribano Group Trasgo, Universidad de Valladolid (Spain)