#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
__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;
}
__global__ void parallel_monte_carlo_integrate(curandStatePhilox4_32_10_t* state, int R, float* result) {
    //srand(time(NULL));
    //float* dev_state;
    //cudaMalloc((void**)&dev_state, sizeof(float));
    //curandStatePhilox4_32_10_t* dev_state;
    float w = 1 / (1.0 * R);
    int success = 0;
    int cp = 0;
    int i = threadIdx.x;
    for (int j = 0; j < pnt; j++) {
        float x = 1.0 * state;
        x = 1.0 * state / RAND_MAX;
        x = 1.0 * state / RAND_MAX + i * w;
        x = 1.0 * state / RAND_MAX / R + i * w;
        float y = 1.0 * state / RAND_MAX;
        success = success + int(f(x) < y);

        //printf("x=%lf\n", x);
        cp++;
    }
    *result = success * 1.0 / cp;
    printf("%lf\n", *result);
}
int main(void) {
    //printf("%d", RAND_MAX);
    float result;
    int R = 2;
    float* dev_result;
    cudaMalloc((void**)&dev_result, sizeof(float));
    //cudaMemcpy(&result, &result_gpu, size_t count, cudaMemcpyHostToDevice = 1);
    parallel_monte_carlo_integrate << <1, R >> > (curandStatePhilox4_32_10_t* state, R, dev_result);
    cudaMemcpy(&result, dev_result, sizeof(float), cudaMemcpyDeviceToHost);
    /*for (int R = 1; R < 100; R++) {
        parallel_monte_carlo_integrate(R, &result);
        printf("%x %lf\n", &result, result);
    }*/
    printf("%lf", result);
    return 0;
}