Scribd
Upload
137 views

Exercises: Parallel Programming Lab

The document describes a parallel programming lab assignment to calculate the standard deviation of a distributed set of random numbers using MPI. The program involves: 1) Generating random numbers on each process. 2) Calculating the local sum and squared differences from the mean on each process. 3) Reducing the local sums to a global sum and squared differences to calculate the mean and standard deviation. 4) Printing the results.

Uploaded by

Muhammad Ali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
137 views

Exercises: Parallel Programming Lab

The document describes a parallel programming lab assignment to calculate the standard deviation of a distributed set of random numbers using MPI. The program involves: 1) Generating random numbers on each process. 2) Calculating the local sum and squared differences from the mean on each process. 3) Reducing the local sums to a global sum and squared differences to calculate the mean and standard deviation. 4) Printing the results.

Uploaded by

Muhammad Ali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

CSL-342: Parallel Programming lab

Semester BSIT – 06
Student Name: Muhammad Faisal
Registration: 43865

Lab 09: Performing Parallel Rank with MPI

Exercises

Exercise 1

Write a program to finding the standard deviation of a distributed set of


numbers. For those that may have forgotten, standard deviation is a measure
of the dispersion of numbers from their mean. A lower standard deviation
means that the numbers are closer together and vice versa for higher standard
deviations.

Code

#include <stdio.h> #include <stdlib.h> #include <stdio.h> #include <stdlib.h>


#include <mpi.h> #include <math.h> #include <mpi.h> #include <assert.h>
#include <assert.h> #include <time.h>
// Creates an array of random numbers. // Creates an array of random numbers.
Each number has a value from 0 - 1 float Each number has a value from 0 - 1 float
*create_rand_nums(int num_elements) { *create_rand_nums(int num_elements) {
float *rand_nums = (float float *rand_nums = (float
*)malloc(sizeof(float) * num_elements); *)malloc(sizeof(float) * num_elements);
assert(rand_nums != NULL); int i; for assert(rand_nums != NULL); int i; for
(i = 0; i < num_elements; i++) { (i = 0; i < num_elements; i++) {
rand_nums[i] = (rand() / rand_nums[i] = (rand() /
(float)RAND_MAX); } return rand_nums; (float)RAND_MAX); } return rand_nums;
} }
int main(int argc, char** argv) { if int main(int argc, char** argv) { if
(argc != 2) { fprintf(stderr, "Usage: (argc != 2) { fprintf(stderr, "Usage:
avg num_elements_per_proc\n"); avg num_elements_per_proc\n");
exit(1); } exit(1); }
int num_elements_per_proc = int num_elements_per_proc =
atoi(argv[1]); atoi(argv[1]);
MPI_Init(NULL, NULL); MPI_Init(NULL, NULL);
int world_rank; int world_rank;
MPI_Comm_rank(MPI_COMM_WORLD, MPI_Comm_rank(MPI_COMM_WORLD,
&world_rank); int world_size; &world_rank); int world_size;
MPI_Comm_size(MPI_COMM_WORLD, MPI_Comm_size(MPI_COMM_WORLD,
&world_size); &world_size);
// Create a random array of elements on // Create a random array of elements on
all processes. all processes.
srand(time(NULL)*world_rank); // Seed the srand(time(NULL)*world_rank); // Seed
random number generator of processes the random number generator to get
uniquely float *rand_nums = NULL; different results each time for each
rand_nums = processor float *rand_nums = NULL;
create_rand_nums(num_elements_per_proc); rand_nums =
// Sum the numbers locally float create_rand_nums(num_elements_per_proc);
local_sum = 0; int i; for (i = 0; i <
num_elements_per_proc; i++) { float local_sum = 0; int i; for (i =
local_sum += rand_nums[i]; } 0; i < num_elements_per_proc; i++) {
// Reduce all of the local sums into local_sum += rand_nums[i]; }
the global sum in order to // calculate // Print the random numbers on each
the mean float global_sum; process printf("Local sum for process
MPI_Allreduce(&local_sum, &global_sum, 1, %d - %f, avg = %f\n",
MPI_FLOAT, MPI_SUM, world_rank, local_sum, local_sum /
MPI_COMM_WORLD); float mean = num_elements_per_proc);
global_sum / (num_elements_per_proc * // Reduce all of the local sums into
world_size); the global sum float global_sum;
// Compute the local sum of the squared MPI_Reduce(&local_sum, &global_sum, 1,
differences from the mean float MPI_FLOAT, MPI_SUM, 0,
local_sq_diff = 0; for (i = 0; i < MPI_COMM_WORLD);
num_elements_per_proc; i++) { // Print the result if (world_rank ==
local_sq_diff += (rand_nums[i] - mean) * 0) { printf("Total sum = %f, avg =
(rand_nums[i] - mean); } %f\n", global_sum, global_sum
// Reduce the global sum of the squared / (world_size * num_elements_per_proc));
differences to the root process // and }
print off the answer float // Clean up free(rand_nums);
global_sq_diff; MPI_Barrier(MPI_COMM_WORLD);
MPI_Reduce(&local_sq_diff, MPI_Finalize(); }
&global_sq_diff, 1, MPI_FLOAT, MPI_SUM,
0, MPI_COMM_WORLD);
if (world_rank == 0) { float stddev =
sqrt(global_sq_diff /
(num_elements_per_proc * world_size));
printf("Mean - %f, Standard deviation =
%f\n", mean, stddev); }
// Clean up free(rand_nums);
MPI_Barrier(MPI_COMM_WORLD);
MPI_Finalize(); }

Output:

You might also like