#include "cuda_runtime.h"
#include "device_launch_parameters.h"
#include <stdio.h>
#include <stdlib.h>
#include <cstdlib>
#include <time.h>
#include <curand_kernel.h>
#define pnt 100
#define SEED 9876543
#define THREADS 80
#define BLOCKS 10
__global__ void mykernel(void) {
}
__global__ void addKernel(int* c, const int* a, const int* b) {
    int i = threadIdx.x;
    c[i] = a[i] + b[i];
}
__device__ float f(float x) {
    return x;
}
// инициализация генератора
__global__ void initfGENS(curandStatePhilox4_32_10_t* d_gen) {
    for (int id = threadIdx.x + blockIdx.x * blockDim.x; id < THREADS * BLOCKS; id += THREADS * BLOCKS) {
        curand_init(SEED, id, 0, d_gen + id);
    }

}
__global__ void parallel_monte_carlo_integrate(curandStatePhilox4_32_10_t* d_gen, int R, float* result) {
    float w = 1 / (1.0 * R);
    int success = 0;
    int cp = 0;
    int id = threadIdx.x + blockIdx.x * blockDim.x;
    for (int j = 0; j < pnt; j++) {
        float x = 1.0 * curand_uniform(&d_gen[id]) / R + w * id;
        float y = 1.0 * curand_uniform(&d_gen[id]);
        success = success + int(f(x) < y);
        cp++;
    }
    *result = success * 1.0 / cp;
    printf("%lf\n", *result);
}
int main() {
    curandStatePhilox4_32_10_t* d_gen;
    cudaMalloc((void**)&d_gen, THREADS * BLOCKS * sizeof(curandStatePhilox4_32_10_t));
    initfGENS << < 1, 10 >> > (d_gen);
    float result;
    int R = 10;
    float* dev_result;
    cudaMalloc((void**)&dev_result, sizeof(float));
    parallel_monte_carlo_integrate << <THREADS, BLOCKS >> > (d_gen, R, dev_result);
    cudaMemcpy(&result, dev_result, sizeof(float), cudaMemcpyDeviceToHost);
    printf("%lf", result);
    //освобождение памяти
    cudaFree(d_gen);
    cudaFree(dev_result);
    return 0;
}