benchmark.cu

#include "cuda_runtime.h"
#include "device_launch_parameters.h"

#include <stdio.h>
#include <stdio.h>
#include <chrono>
#include <iostream>

__global__ void vectorized_elementwise_kernel(int N) {

}

inline void benchmarkTest(int iterations)
{
    double elapsed_time = 0;
    for (int i = 0; i < iterations; i++)
    {
        auto start = std::chrono::steady_clock::now();
        vectorized_elementwise_kernel
            << <8, 128, 0, nullptr >> > (4000);
        auto end = std::chrono::steady_clock::now();
        elapsed_time += std::chrono::duration<double>(end - start).count();
    }
    std::cout << "benchmark iterations " << iterations << " took " << elapsed_time << "\n";
}

int main()
{
    cudaError_t cudaStatus;
    cudaStatus = cudaSetDevice(0);
    benchmarkTest(750);
    return 0;

}