Reverse Array

1
#include <cuda_runtime.h>
2

3
__global__ void reverse_array(float* input, int N) {
4
    int idx = blockDim.x * blockIdx.x + threadIdx.x;
5
    if (idx < N / 2) {
6
        float tmp = input[idx];
7
        input[idx] = input[N - 1 - idx];
8
        input[N - 1 - idx] = tmp;
9
    }
10
}
11

12
// input is device pointer
13
extern "C" void solve(float* input, int N) {
14
    int threadsPerBlock = 256;
15
    int blocksPerGrid = (N + threadsPerBlock - 1) / threadsPerBlock;
16

17
    reverse_array<<<blocksPerGrid, threadsPerBlock>>>(input, N);
18
    cudaDeviceSynchronize();
19
}