#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include <cuda_runtime.h>


void *__gxx_personality_v0; //c++ shenanigans

extern void init(int n, int *res);
extern void end();
extern void launch(float *a, uint64_t l, float r);


int main(int argc, char *argv[]) {
    int out = 0;
    if (argc <= 2) {
        fprintf(stderr, "testDRP <vector length(int)> <dropout  rate(float)>\n");
        return -1;
    }
    int num = atoi(argv[1]);
    if (num <= 0) {
        fprintf(stderr, "testDRP <vector length(int)> <dropout  rate(float)>\n");
        return -1;
    }
    float rate = strtof(argv[2], NULL);
    if (rate <= 0.0f) {
        fprintf(stderr, "testDRP <vector length(int)> <dropout  rate(float)>\n");
        return -1;
    }
    int res;
    init(num, &res);
    if (!res) {
        fprintf(stderr, "Failed to initialize curand states.\n");
        return -1;
    }


    float *a = 0;
    float *b = 0;
    b = malloc(num*sizeof(float));
    if (!b) {
        fprintf(stderr, "Host memory error\n");
        out = -1;
        goto exit;
    }
    for (int i = 0; i < num; i++) {
        b[i] = (float)rand() / (float)RAND_MAX;
    }

    if (cudaSuccess != cudaMalloc((void**)&a, num*sizeof(float))) {
        fprintf(stderr, "Device memory error\n");
        out = -1;
        goto exit;
    }

    cudaMemcpy(a, b, num * sizeof(float), cudaMemcpyHostToDevice);
    memset(b, 0, num*sizeof(float));
    launch(a, num, rate);
    cudaMemcpy(b, a, num * sizeof(float), cudaMemcpyDeviceToHost);

    for (int i = 0; i < num; i++) {
        if (!b[i])
            fprintf(stdout, "0\n");
    }

    exit:
        cudaFree(a);
        free(b);
        end(); //Free globalstate array
        return out;
}