KMP-MPI/src/parallel.c

#include <stdio.h>
#include <string.h>
#include "kmp.h"
#include "util.h"
#include <mpi.h>
#include <malloc.h>

#ifdef TIME
#include <papi.h>
#endif

#define MASTER 0
#define DEFAULT_TAG 17

void find_end(int, char *, char *, int **, int *);

void search_for_splitted_pattern(int *residue, int *match_number, int **matches);

void calculate_absolute_indices(int match_number, int *matches);

void collect_results(int *match_number, int *matches);

int rank, size;
int text_len;       // length of all text
int private_text_len;   // length of private text
int pattern_len;    //length of the pattern
char *text;
char *pattern;
int remain = 0;
int *text_piece;
int *displacements;
char *private_text;
int *match_numbers;
int *total_matches;
int total_match_number;

int main(int argc, char **argv) {
    MPI_Init(NULL, NULL);

    MPI_Comm_size(MPI_COMM_WORLD, &size);
    MPI_Comm_rank(MPI_COMM_WORLD, &rank);

#ifdef TIME
    long_long start_time = PAPI_get_real_usec();
#endif

    char *text_file_path = get_text_file(argc, argv);
    private_text = read_file_portion(text_file_path, size, rank, &private_text_len, &remain);
    pattern = read_file("data/pattern.txt", &pattern_len);

#ifdef LOG
    printf("%d -> input: %s\n", rank, private_text);
    printf("%d -> text len: %d\n", rank, private_text_len);
#endif

    // create and build lps array
    int lps[pattern_len];
    create_lps(pattern, pattern_len, lps);

    // search for matches
    int match_number = 0;
    int residue;
#ifdef TIME
    long_long start_search = PAPI_get_real_usec();
#endif
    int *matches = search_pattern(private_text, pattern, lps, &match_number, &residue);

    search_for_splitted_pattern(&residue, &match_number, &matches);
#ifdef TIME
    long_long stop_search = PAPI_get_real_usec();
#endif

    //free(private_text);

    calculate_absolute_indices(match_number, matches);

#ifdef LOG
    printf("%d -> result number: %d\n", rank, match_number);
#endif

#ifdef TIME
    long_long start_collect = PAPI_get_real_usec();
#endif

    collect_results(&match_number, matches);

#ifdef TIME
    long_long end_time = PAPI_get_real_usec();
#endif

    MPI_Finalize();

    if (rank == MASTER) {
        printf("total matches: %d\n", total_match_number);
#ifdef LOG
        printf("matches index: ");
        print_array(total_matches, total_match_number);
#endif

#ifdef TIME
        printf("total elapsed: %d\n", end_time - start_time);
        printf("search elapsed: %d\n", stop_search - start_search);
        printf("collect elapsed: %d\n", end_time - start_collect);
#endif
    }
}

void find_end(int residue, char *pattern, char *text, int **matches, int *match_number) {
    int pattern_index = residue;
    int text_index = 0;
    int text_len = strlen(text);
    int pattern_len = strlen(pattern);

    if (pattern_len - residue > text_len) return;

    while (pattern_index < pattern_len && pattern[pattern_index] == text[text_index]) {
        pattern_index++;
        text_index++;
    }

    if (pattern_index != pattern_len) return;

    *match_number = *match_number + 1;
    *matches = (int *) realloc(*matches, *match_number * sizeof(int));
    (*matches)[*match_number - 1] = -residue;
}

void search_for_splitted_pattern(int *residue, int *match_number, int **matches) {
    // send the residue to the next process
    if (rank + 1 != size) {
        MPI_Send(residue, 1, MPI_INT, rank + 1, DEFAULT_TAG, MPI_COMM_WORLD);
    }

    // receiving the residue from the previous process
    if (rank != 0) {
        MPI_Recv(residue, 1, MPI_INT, rank - 1, DEFAULT_TAG, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
        if (*residue != 0)
            // search for a split match
            find_end(*residue, pattern, private_text, matches, match_number);
    }

}

void calculate_absolute_indices(int match_number, int *matches) {
    // transformation of match indices from relative to absolute position.
    int shift = rank * (private_text_len - remain); // the rest is zero for all but the last process
    apply_shift(shift, matches, match_number);
}

void collect_results(int *match_number, int *matches) {
    if (rank == MASTER) {
        // allocation for the array containing the number of matches of each process
        match_numbers = (int *) malloc(sizeof(int) * size);
    }

    // collection of the number of matches from each process
    MPI_Gather(match_number, 1, MPI_INT, match_numbers, 1, MPI_INT, MASTER, MPI_COMM_WORLD);

    if (rank == MASTER) {
        // preparation of the data structures needed to receive match indices from all processes
        total_match_number = sum_array(match_numbers, size);
        total_matches = (int *) malloc(sizeof(int) * total_match_number);

        displacements = (int *) malloc(sizeof(int) * size);
        displacements[0] = 0;
        for (int i = 0; i < size - 1; ++i) {
            displacements[i + 1] = displacements[i] + match_numbers[i];
        }
    }

    // collection of match indices from each process
    // gatherv is necessary because each process will have a different number of matches
    MPI_Gatherv(
            matches,
            *match_number,
            MPI_INT,
            total_matches,
            match_numbers,
            displacements,
            MPI_INT, MASTER,
            MPI_COMM_WORLD
    );
}