The purpose of the project is to multiply two matrices in the size of 1K to 5K in 3 different ways and analyze the effects on performance.
The multiplication of two matrices is to be implemented as:
- a serial program
- an OpenMP shared memory program (parallel)
- a loop blocking method program
You can install the application with clone this repository by the commands given below
# clone the repository
$ git clone https://github.com/selenkutanoglu/Matrix-Multiplication
Matix size, value type of the matrix and the block size of for looping part defined as macro variables in functions.h file.
#define SIZE 1000 //Set the size of matrix
#define type double //Set the type of matrix (double - float)
#define block_size 10 //System block size (optimum value as a result of tests)
. bFunctions.cpp loop blocing program functions
. functions.h whole funtions declarations, matrix parameters and libraries needed
. main.cpp main program
. pFunctions.cpp OpenMP parallel program functions
. README.md This README file you're looking
. sFunctions.cpp serial program functions
To compile the serial program, sFunctions.cpp and main.cpp will be to compiled with g++ and object file named as serial.
$ g++ -o serial sFunctions.cpp main.cpp
$ ./serial
To compile the OpenMP parallel program, pFunctions.cpp and main.cpp will be to compiled with g++ (with -fopenmp parameter) and object file named as parallel.
$ g++ -O -Wall -std=c++0x -g -fopenmp -o parallel pFunctions.cpp main.cpp
$ ./parallel
To compile the loop blocking program, bFunctions.cpp and main.cpp will be to compiled with g++ and object file named as block.
$ g++ -o block bFunctions.cpp main.cpp
$ ./block