Practical 1

#include <iostream>
#include <vector>
#include <queue>
#include <omp.h>
using namespace std;
// Graph class representing the adjacency list
class Graph {
int V; // Number of vertices
vector<vector<int>> adj; // Adjacency list
public:
Graph(int V) : V(V), adj(V) {}
// Add an edge to the graph
void addEdge(int v, int w) {
adj[v].push_back(w);
}
// Parallel Depth-First Search
void parallelDFS(int startVertex) {
vector<bool> visited(V, false);
parallelDFSUtil(startVertex, visited);
}
// Parallel DFS utility function
void parallelDFSUtil(int v, vector<bool>& visited) {
visited[v] = true;
cout << v << " ";
#pragma omp parallel for
for (int i = 0; i < adj[v].size(); ++i) {
int n = adj[v][i];
if (!visited[n])
parallelDFSUtil(n, visited);
}
}
// Parallel Breadth-First Search
void parallelBFS(int startVertex) {
vector<bool> visited(V, false);
queue<int> q;
visited[startVertex] = true;
q.push(startVertex);
while (!q.empty()) {
int v = q.front();
q.pop();
cout << v << " ";
#pragma omp parallel for
for (int i = 0; i < adj[v].size(); ++i) {
int n = adj[v][i];
if (!visited[n]) {
 visited[n] = true;
q.push(n);
}
}}}};
int main() {
// Create a graph
Graph g(7);
g.addEdge(0, 1);
g.addEdge(0, 2);
g.addEdge(1, 3);
g.addEdge(1, 4);
g.addEdge(2, 5);
g.addEdge(2, 6);
cout << "Depth-First Search (DFS): ";
g.parallelDFS(0);
cout << endl;
cout << "Breadth-First Search (BFS): ";
g.parallelBFS(0);
cout << endl;
return 0;
}
Practical 2 (Parallel Bubble Sort)

#include<iostream>
#include<omp.h>
using namespace std;
void bubble(int array[], int n){
for (int i = 0; i < n - 1; i++){
for (int j = 0; j < n - i - 1; j++){
if (array[j] > array[j + 1]) swap(array[j], array[j + 1]);
}
}
}
void pBubble(int array[], int n){
//Sort odd indexed numbers
for(int i = 0; i < n; ++i){
#pragma omp for
for (int j = 1; j < n; j += 2){
if (array[j] < array[j-1])
{
swap(array[j], array[j - 1]);
}
}
// Synchronize
#pragma omp barrier
//Sort even indexed numbers
#pragma omp for
for (int j = 2; j < n; j += 2){
if (array[j] < array[j-1])
{
swap(array[j], array[j - 1]);
}
}
}
}
void printArray(int arr[], int n){
for(int i = 0; i < n; i++) cout << arr[i] << " ";
cout << "\n";
}
int main(){
// Set up variables
int n = 10;
int arr[n];
int brr[n];
double start_time, end_time;
// Create an array with numbers starting from n to 1
 for(int i = 0, j = n; i < n; i++, j--) arr[i] = j;
// Sequential time
start_time = omp_get_wtime();
bubble(arr, n);
end_time = omp_get_wtime();
cout << "Sequential Bubble Sort took : " << end_time - start_time << " seconds.\n";
printArray(arr, n);
// Reset the array
for(int i = 0, j = n; i < n; i++, j--) arr[i] = j;
// Parallel time
start_time = omp_get_wtime();
pBubble(arr, n);
end_time = omp_get_wtime();
cout << "Parallel Bubble Sort took : " << end_time - start_time << " seconds.\n";
printArray(arr, n);
}
Practical 2 (Parallel Merge Sort)
#include <iostream>
#include <omp.h>
using namespace std;
void merge(int arr[], int low, int mid, int high) {
// Create arrays of left and right partititons
int n1 = mid - low + 1;
int n2 = high - mid;
int left[n1];
int right[n2];
// Copy all left elements
for (int i = 0; i < n1; i++) left[i] = arr[low + i];
// Copy all right elements
for (int j = 0; j < n2; j++) right[j] = arr[mid + 1 + j];
// Compare and place elements
int i = 0, j = 0, k = low;
while (i < n1 && j < n2) {
if (left[i] <= right[j]){
arr[k] = left[i];
i++;
}
else{
arr[k] = right[j];
j++;
}
k++;
}
// If any elements are left out
while (i < n1) {
arr[k] = left[i];
i++;
k++;
}
while (j < n2) {
arr[k] = right[j];
j++;
k++;
}
}
void parallelMergeSort(int arr[], int low, int high) {
if (low < high) {
int mid = (low + high) / 2;
#pragma omp parallel sections
{
#pragma omp section
{
parallelMergeSort(arr, low, mid);
}
#pragma omp section
{ parallelMergeSort(arr, mid + 1, high);
 }
}
merge(arr, low, mid, high);
}
}
void mergeSort(int arr[], int low, int high) {
if (low < high) {
int mid = (low + high) / 2;
mergeSort(arr, low, mid);
mergeSort(arr, mid + 1, high);
merge(arr, low, mid, high);
}
}
int main() {
int n = 10;
int arr[n];
double start_time, end_time;
// Create an array with numbers starting from n to 1.
for(int i = 0, j = n; i < n; i++, j--) arr[i] = j;
// Measure Sequential Time
start_time = omp_get_wtime();
mergeSort(arr, 0, n - 1);
end_time = omp_get_wtime();
cout << "Time taken by sequential algorithm: " << end_time - start_time << " seconds\n";
// Reset the array
for(int i = 0, j = n; i < n; i++, j--) arr[i] = j;
//Measure Parallel time
start_time = omp_get_wtime();
parallelMergeSort(arr, 0, n - 1);
end_time = omp_get_wtime();
cout << "Time taken by parallel algorithm: " << end_time - start_time << " seconds";
return 0;
}
Practical 3

#include<iostream>
#include<omp.h>
using namespace std;
int minval(int arr[], int n){
int minval = arr[0];
#pragma omp parallel for reduction(min : minval)
for(int i = 0; i < n; i++){
if(arr[i] < minval) minval = arr[i];
}
return minval;
}
int maxval(int arr[], int n){
int maxval = arr[0];
#pragma omp parallel for reduction(max : maxval)
for(int i = 0; i < n; i++){
if(arr[i] > maxval) maxval = arr[i];
}
return maxval;
}
int sum(int arr[], int n){
int sum = 0;
#pragma omp parallel for reduction(+ : sum)
for(int i = 0; i < n; i++){
sum += arr[i];
}
return sum;
}
int average(int arr[], int n){
return (double)sum(arr, n) / n;
}
int main(){
int n = 5;
int arr[] = {1,2,3,4,5};
cout << "The minimum value is: " << minval(arr, n) << '\n';
cout << "The maximum value is: " << maxval(arr, n) << '\n';
cout << "The summation is: " << sum(arr, n) << '\n';
cout << "The average is: " << average(arr, n) << '\n';
return 0;
}
4/15/24, 12:19 PM Matrix_Multiplication.ipynb - Colab

keyboard_arrow_down Practical no 4(Matrix Multiplication)

# Set up CUDA
#First Change runtime to GPU and run this cell
!pip install git+https://github.com/thefaizan1/Cuda.git
%load_ext nvcc_plugin

Collecting git+https://github.com/thefaizan1/Cuda.git
Cloning https://github.com/thefaizan1/Cuda.git to /tmp/pip-req-build-hdaihoaq
Running command git clone --filter=blob:none --quiet https://github.com/thefaizan1/Cuda.git /tmp/pip-req-build-hdaihoaq
Resolved https://github.com/thefaizan1/Cuda.git to commit 7a67b82da3d96cd6fbb7c8f8d6044f601a6740f2
Preparing metadata (setup.py) ... done
The nvcc_plugin extension is already loaded. To reload it, use:
%reload_ext nvcc_plugin

%%cu
#include<iostream>
using namespace std;
__global__ void multiply(int* A, int* B, int* C, int size) {
// Uses thread idices and block indices to compute each element
int row = blockIdx.y * blockDim.y + threadIdx.y;
int col = blockIdx.x * blockDim.x + threadIdx.x;
if (row < size && col < size) {
int sum = 0;
for (int i = 0; i < size; i++) {
sum += A[row * size + i] * B[i * size + col];
}
C[row * size + col] = sum;
}
}
void initialize(int* matrix, int size) {
for (int i = 0; i < size * size; i++) {
matrix[i] = rand() % 10;
}
}
void print(int* matrix, int size) {
for (int row = 0; row < size; row++) {
for (int col = 0; col < size; col++) {
cout << matrix[row * size + col] << " ";
}
cout << '\n';
}
cout << '\n';
}
int main() {
int* A, * B, * C;
int N = 2;
int blockSize = 16;
int matrixSize = N * N;
size_t matrixBytes = matrixSize * sizeof(int);
A = new int[matrixSize];
B = new int[matrixSize];
C = new int[matrixSize];
initialize(A, N);
initialize(B, N);
cout << "Matrix A: \n";
print(A, N);
cout << "Matrix B: \n";
print(B, N);
int* X, * Y, * Z;
cudaMalloc(&X, matrixBytes);
cudaMalloc(&Y, matrixBytes);
cudaMalloc(&Z, matrixBytes);
// Copy values from A to X
cudaMemcpy(X, A, matrixBytes, cudaMemcpyHostToDevice);
// Copy values from A to X and B to Y
cudaMemcpy(Y, B, matrixBytes, cudaMemcpyHostToDevice);
// Threads per CTA dimension
int THREADS = 2;
// Blocks per grid dimension (assumes THREADS divides N evenly)
int BLOCKS = N / THREADS;
// Use dim3 structs for block and grid dimensions
dim3 threads(THREADS, THREADS);
dim3 blocks(BLOCKS, BLOCKS);
// Launch kernel
multiply<<<BLOCKS, THREADS>>>(X, Y, Z, N);
cudaMemcpy(C, Z, matrixBytes, cudaMemcpyDeviceToHost);
cout << "Multiplication of matrix A and B: \n";
print(C, N);

https://colab.research.google.com/drive/1NWYvOrr3cEvq8QOxknyCpLKn1uwaLLoA#scrollTo=DKEzDjZDgHyl&printMode=true 1/2
4/15/24, 12:19 PM Matrix_Multiplication.ipynb - Colab
p ( , );
delete[] A;
delete[] B;
delete[] C;
cudaFree(X);
cudaFree(Y);
cudaFree(Z);
return 0;
}

Output
output Matrix A:
3 6
7 5

Matrix B:
3 5
6 2

Multiplication of matrix A and B:

45 27
0 0

https://colab.research.google.com/drive/1NWYvOrr3cEvq8QOxknyCpLKn1uwaLLoA#scrollTo=DKEzDjZDgHyl&printMode=true 2/2
4/15/24, 12:11 PM Vector_Addition.ipynb - Colab

keyboard_arrow_down Practical No 4 (Vector Addition)

# Set up CUDA
#First Change runtime to GPU and run this cell
!pip install git+https://github.com/thefaizan1/Cuda.git
%load_ext nvcc_plugin

Collecting git+https://github.com/thefaizan1/Cuda.git
Cloning https://github.com/thefaizan1/Cuda.git to /tmp/pip-req-build-0q6io3kp
Running command git clone --filter=blob:none --quiet https://github.com/thefaizan1/Cuda.git /tmp/pip-req-build-0q6io3kp
Resolved https://github.com/thefaizan1/Cuda.git to commit 7a67b82da3d96cd6fbb7c8f8d6044f601a6740f2
Preparing metadata (setup.py) ... done
The nvcc_plugin extension is already loaded. To reload it, use:
%reload_ext nvcc_plugin

%%cu

#include <iostream>
using namespace std;

__global__
void add(int* A, int* B, int* C, int size) {
int tid = blockIdx.x * blockDim.x + threadIdx.x;

if (tid < size) {

C[tid] = A[tid] + B[tid];
}
}

void initialize(int* vector, int size) {

for (int i = 0; i < size; i++) {
vector[i] = rand() % 10;
}
}

void print(int* vector, int size) {

for (int i = 0; i < size; i++) {
cout << vector[i] << " ";
}
cout << endl;
}

int main() {
int N = 50;
int* A, * B, * C;

int vectorSize = N;
size_t vectorBytes = vectorSize * sizeof(int);

A = new int[vectorSize];
B = new int[vectorSize];
C = new int[vectorSize];

initialize(A, vectorSize);
initialize(B, vectorSize);

cout << "Vector A: ";

print(A, N);
cout << "Vector B: ";
print(B, N);

int* X, * Y, * Z;
cudaMalloc(&X, vectorBytes);
cudaMalloc(&Y, vectorBytes);
cudaMalloc(&Z, vectorBytes);

cudaMemcpy(X, A, vectorBytes, cudaMemcpyHostToDevice);

cudaMemcpy(Y, B, vectorBytes, cudaMemcpyHostToDevice);

int threadsPerBlock = 256;

int blocksPerGrid = (N + threadsPerBlock - 1) / threadsPerBlock;

add<<<blocksPerGrid, threadsPerBlock>>>(X, Y, Z, N);

cudaMemcpy(C, Z, vectorBytes, cudaMemcpyDeviceToHost);

cout << "Addition: ";

https://colab.research.google.com/drive/1LDxNjqvdLg2jRsQ2ZlcJW9tAJ-Jzo9dM#scrollTo=eCRZ0p3urPBl&printMode=true 1/2
4/15/24, 12:11 PM Vector_Addition.ipynb - Colab
;
print(C, N);

delete[] A;
delete[] B;
delete[] C;

cudaFree(X);
cudaFree(Y);
cudaFree(Z);

return 0;
}

//Output
output Vector A: 3 6 7 5 3 5 6 2 9 1 2 7 0 9 3 6 0 6 2 6 1 8 7 9 2 0 2 3 7 5 9 2 2 8 9 7 3 6 1 2 9 3 1 9 4 7 8 4 5 0
Vector B: 3 6 1 0 6 3 2 0 6 1 5 5 4 7 6 5 6 9 3 7 4 5 2 5 4 7 4 4 3 0 7 8 6 8 8 4 3 1 4 9 2 0 6 8 9 2 6 6 4 9
Addition: 6 12 8 5 9 8 8 2 15 2 7 12 4 16 9 11 6 15 5 13 5 13 9 14 6 7 6 7 10 5 16 10 8 16 17 11 6 7 5 11 11 3 7 17 13 9 14 10 9 9

https://colab.research.google.com/drive/1LDxNjqvdLg2jRsQ2ZlcJW9tAJ-Jzo9dM#scrollTo=eCRZ0p3urPBl&printMode=true 2/2