// DPC++ material used in the code sample:
// •	A one dimensional array of data shared between CPU and offload device.
// •	A device queue and kernel.
#include <CL/sycl.hpp>
#include <array>
#include <iostream>
#include <windows.h>

using namespace sycl;

cpu_selector d_selector;
queue q(d_selector);

unsigned long long  DeltaFPS, CountFPS;

constexpr size_t array_size = 100 * 100 * 100;

float4* a            = malloc_shared<float4>(array_size, q);
float4* sum_parallel = malloc_shared<float4>(array_size, q);

void    VectorAdd(queue &q, const float4* a, float4* sum, size_t size)
{
    range<1> num_items{size};
    q.parallel_for(num_items, [=](auto i)
    {
        sum[i] = a[i] + 10;
    }).wait();
}

void    ShowFPS()
{
    CountFPS++;
    if (GetTickCount64() > DeltaFPS + 1000)
    {
        printf("FPS = %lld\n", CountFPS);

        CountFPS = 0;
        DeltaFPS = GetTickCount64();
    }
}
int     main()
{
    std::cout << "Running on device: " << q.get_device().get_info<info::device::name>() << "\n";

    while (true)
    {
        VectorAdd(q, a, sum_parallel, array_size);
        ShowFPS();
    }

    return 0;
}