#include "mpi.h"
#include <fcntl.h>
#include <stdlib.h>     //the standard library for c
#include <sys/time.h>   //used to have access to gettimeofday to calculate time taken
#include <time.h>       //used to have access to gettimeofday to calculate time taken
#include <stdio.h>      //used for printf command to allow easy printing
const int filter_length = 1024;                       //length of the filter
const int trace_count = 1024;                         //number of arrays in the trace
const int trace_length = 16384;                       //length of each trace
//place items in heap beacuse the stack is not large enough for them
float d[1024][17408] = {{0}};   //holds the data values that will be filtered.
                                //This includes 512 0 value buffers surrounding the data
           //holds the data that will be used to filter the data matrix
float out[1024][16384] = {{0}};   //holds the filtered output
float fb[(sizeof(float)* filter_length)];
float data[trace_count][trace_length + filter_length] = {{0}};


void performOperation(int start, int final);

int main(int argc, char** argv) {

    int currentRow = 0;
    int dataFD = open("/home/data/files/data.bin", O_RDONLY);
    for (currentRow = 0; currentRow < trace_count; currentRow++) {      //for each row
        off_t offset = currentRow * trace_length * sizeof(float);       //calculate offset
        lseek(dataFD, offset, SEEK_SET);                                       //seek to offset in the code
        read(dataFD, &data[currentRow][512], sizeof(float) * trace_length);   //read next 16k elements
    }

    //reading Filter from flt.bin
    int fd = open("/home/data/files/filt.bin", O_RDONLY);
    read(fd,fb,sizeof(float)*filter_length);
    int rank;                   // Process order
    int size;                   // Number of processes

    int i, j;
    // for(i = 0; i < trace_count;i++) {
    //     for(j = 512; j < trace_length+512; j++){
    //         printf("data[%d][%d] = %f\n", i, j, data[i][j]);
    //     }
    // }
    MPI_Init (&argc, &argv);
    MPI_Comm_size (MPI_COMM_WORLD, &size);
    MPI_Comm_rank (MPI_COMM_WORLD, &rank);

    int count = 0;
    double begin = MPI_Wtime();

    int start = ((rank * trace_count) / size);
    int final = (((rank+1) * trace_count) / size);
    performOperation(start, final);

    MPI_Barrier(MPI_COMM_WORLD);

    begin -= MPI_Wtime();
    begin *= -1;
    MPI_Finalize ();
    if(rank==0) {
       printf("Execution across %d nodes took %f seconds.\n", size, begin);
    }

    // for(i = 0; i < trace_count;i++) {
    //     for(j = 0; j < trace_length; j++){
    //         printf("out[%d][%d] = %f\n", i, j, out[i][j]);
    //     }
    // }
    return 0;

}

void performOperation(int start, int final) {
    int j = start;
    int i = 0;
    int k = 0;
    int count = 1;
    int diff = final - start;

    for( j=start; j< final; j++ ) { //iterate over each trace
        for( i=0; i<trace_length; i++ ) {
            for( k=0; k<filter_length; k++ ) {
                out[j][i] += fb[k] * data[j][i+k - (filter_length / 2) + 512];
            }
        }
    }
}