Cannot send the result back from GPU - Windows TDR event

Question

I coded a little program "Finding Prime Numbers" as my practice. After I tried to set more blocks successfully in GPU. I encountered another problem that I cannot understand. I found that the result cannot send back to the host when I input that I would like the program to count all the prime numbers which are less than 200,000. It showed memory failed message. According to the comments and some materials, I think I met Windows TDR event. My Screen flashed and the message "Display driver stopped response and recovered" came out when I run this program. What should I solve this problem?

My code is as below:

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

#include <iostream>
#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <vector>

using namespace std;

cudaError_t findPrimeWithCuda(bool *c, int *a, unsigned int size);

__host__ __device__ bool checkPrime(int i)
{
    for (int m = 2; m <= i - 1; m++)
    {
        if (i%m == 0) return true;
    }
    return false;
}

__global__ void getPrimeKernel(bool *c, int *a, int size)
{
    int i = blockDim.x * blockIdx.x + threadIdx.x;
    if (i >= size) return;
    c[i] = checkPrime(a[i]);
}

void cudaGetPrime(int i)
{
    i = i - 3;
    int *arr = (int *)malloc((size_t)(i * sizeof(int)));
    bool *rst = (bool *)malloc((size_t)(i * sizeof(bool)));
    for (int j = 0; j <= i; j++) arr[j] = j + 3;
    cudaError_t cudaStatus = findPrimeWithCuda(rst, arr, i);
    if (cudaStatus != cudaSuccess) fprintf(stderr,"findPrimeWithCuda failed!!");
}

void w_CudaArray(int lastNum)
{
    time_t t1 = time(NULL);
    cudaGetPrime(lastNum);
    time_t t2 = time(NULL);
    printf("Time spends %d seconds\n", t2 - t1);
}

int main()
{
    int lastNum = 0;
    cout << "Find all prime numbers less than ? ";
    cin >> lastNum;
    w_CudaArray(lastNum);
}

cudaError_t findPrimeWithCuda(bool *c, int *a, unsigned int size)
{
    int *dev_a = 0;
    bool *dev_c = false;
    cudaError_t cudaStatus;

    cudaStatus = cudaSetDevice(0);
    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "cudaSetDevice failed!!");
        goto Error;
    }

    size_t totalm, freem;
    double free_m, total_m, used_m;
    cudaMemGetInfo(&freem, &totalm);
    free_m = (size_t)freem / 1048576.0;
    total_m = (size_t)totalm / 1048576.0;
    used_m = total_m - free_m;
    cout << "Total memory = " << total_m << " MB" << endl;
    cout << "Used memory  = " << used_m << " MB" << endl;
    cout << "Free memory  = " << free_m << " MB" << endl;

    cudaStatus = cudaMalloc((void**)&dev_a, size * sizeof(int));
    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "cudaMalloc dev_a failed!!");
        goto Error;
    }

    cudaStatus = cudaMalloc((void**)&dev_c, size * sizeof(bool));
    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "cudaMalloc dev_c failed!!");
        goto Error;
    }

    cudaStatus = cudaMemcpy(dev_a, a, size * sizeof(int), cudaMemcpyHostToDevice);
    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "cudaMemcpy dev_a failed!!");
        goto Error;
    }

    cudaDeviceProp myCUDA;
    if (cudaGetDeviceProperties(&myCUDA, 0) == cudaSuccess)
    {
        printf("Using device %d:\n", 0);
        printf("%s; global mem: %dB; compute v%d.%d; clock: %d kHz\n",
            myCUDA.name, (int)myCUDA.totalGlobalMem, (int)myCUDA.major,
            (int)myCUDA.minor, (int)myCUDA.clockRate);
    }

    int threadsPerBlock = myCUDA.maxThreadsPerBlock;
    int blocksPerGrid = (size + threadsPerBlock - 1) / threadsPerBlock;
    cout << "Maxium number per block = " << threadsPerBlock << endl;
    cout << "Blocks per Grid = " << blocksPerGrid << endl;

    getPrimeKernel<<<blocksPerGrid, threadsPerBlock>>>(dev_c, dev_a, size);

    cudaStatus = cudaGetLastError();
    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "getPrimeKernel launch failed!!: %s\n",cudaGetErrorString(cudaStatus));
        goto Error;
    }

    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "cudaDeviceSynchorinze returned error code %d after launching getPrimeKernel!\n", cudaStatus);
        goto Error;
    }

    cudaStatus = cudaMemcpy(c, dev_c, size * sizeof(bool), cudaMemcpyDeviceToHost);

    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "cudaMemory failed!");
        goto Error;
    }

    int trueNumber = 0;
    for (int i = 0; i < size; i++)
    {
        if (c[i] == false) trueNumber++;
    }

    cout << "There are " << trueNumber + 2 << " prime numbers!!" << endl;

    cudaStatus = cudaDeviceReset();
    if (cudaStatus != cudaSuccess)
    {
        fprintf(stderr, "cudaDeviceReset failed!!");
    }

Error:
    cudaFree(dev_c);
    cudaFree(dev_a);

    return cudaStatus;
}

After the execution of cudaMemcpy function. I got the error message "cudaMemory failed!"

And my hardware information which I inquired by "cudaDeviceProp" is as below.

================================
Total memory = 1024 MB
Used memory  = 189.293 MB
Free memory  = 834.707 MB
Using device 0:
GeForce GTX 550 Ti; global mem: 1073741824B; compute v2.1; clock: 1800000 kHz
Maxium number per block = 1024
Blocks per Grid = 196
cudaMemory failed!findPrimeWithCuda failed!!