//HOW TO RUN IT ON MY TERMINAL (MAC)
//export LDFLAGS="-L/opt/homebrew/opt/libomp/lib"
//export CPPFLAGS="-I/opt/homebrew/opt/libomp/include"
//clang++ -std=c++17 -O2 -Xpreprocessor -fopenmp $CPPFLAGS main.cpp mylib.cpp -L/opt/homebrew/opt/libomp/lib -lomp -o dotprod
//./dotprod

#include "check_env.h"
#include "mylib.h"
#include <cstdlib>          // atoi()
#include <cstring>          // strncmp()
#include <ctime>
#include <iostream>
#include <omp.h>            // OpenMP
#include <sstream>
#include <string>
using namespace std;

int main(int argc, char const *argv[])
{
    unsigned int N = 5000001;   // smaller default for testing
    int const NLOOPS = 5;       // smaller loop for quick test

    //##########################################################################
    // Read Parameter from command line
    cout << "=== Checking command line parameters ===" << endl;
    for (int i = 1; i < argc; i++)
    {
        string ss(argv[i]);
        if ("-n"==ss && i + 1 < argc)
            N = static_cast<unsigned int>(atoi(argv[i + 1]));
        else
            cout << "Corect call: " << argv[0] << " -n  <number>\n";
    }
    cout << "\nVector size N = " << N << endl;

    //##########################################################################
    check_env(argc, argv);

    //##########################################################################
    cout << "\n=== Starting OpenMP test ===" << endl;
    int nthreads;
    #pragma omp parallel default(none) shared(cout,nthreads)
    {
        int const th_id  = omp_get_thread_num();
        int const nthrds = omp_get_num_threads();
        #pragma omp critical
        cout << "Hello from thread " << th_id << " out of " << nthrds << endl;
        #pragma omp master
        nthreads = nthrds;
    }
    cout << "Total threads started: " << nthreads << endl;

    //##########################################################################
    cout << "\n=== Memory allocation ===" << endl;
    vector<double> x(N), y(N);
    cout << "Allocated memory for vectors x and y (" << N << " doubles each)" << endl;

    //##########################################################################
    cout << "\n=== Data initialization ===" << endl;
    for (unsigned int i = 0; i < N; ++i)
    {
        x[i] = i + 1;
        y[i] = 1.0 / x[i];
    }
    cout << "Vectors initialized: x[i] = i+1, y[i] = 1/x[i]" << endl;

    //##########################################################################
    cout << "\n=== Start Benchmarking inner product ===" << endl;
    double tstart = omp_get_wtime();
    double sk(0.0);
    for (int i = 0; i < NLOOPS; ++i)
    {
        sk = scalar(x, y);
        sk = scalar_trans(x, y);
    }
    double t1 = (omp_get_wtime() - tstart) / NLOOPS;
    cout << "<x, y> = " << sk << endl;
    if (static_cast<unsigned int>(sk) != N)
        cout << "!! WRONG result !!" << endl;

    cout << "Timing (average per loop) in sec: " << t1 << endl;
    cout << "GFLOPS: " << 2.0 * N / t1 / 1024 / 1024 / 1024 << endl;
    cout << "Memory throughput (GiByte/s): " << 2.0 * N / t1 / 1024 / 1024 / 1024 * sizeof(x[0]) << endl;

    //##########################################################################
    cout << "\n=== Testing reduction_vec (combining vectors across threads) ===" << endl;
    cout << "Each thread initializes a vector of size 100; then vectors are combined." << endl;
    auto vr = reduction_vec(100);
    cout << "Resulting combined vector:" << endl;
    cout << vr << endl;

    cout << "\n=== Program finished ===" << endl;
    return 0;
}