我正在尝试使用 opencl 将图像转换为灰度。但是当我尝试创建读缓冲区时。即使缓冲区大小完全在设备的内存限制之内,我也会遇到分段错误。我试过宽度 = 4000 和高度 = 6000 这是我的 opencl 函数。 段错误出现在这一行
cl::Buffer input_buffer(context, CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR, sizeof(uchar) * width * height * 3, input, &err);
void executeOpenCL(uchar*** input, uchar** output, const int height, const int width)
{
cl_int err;
std::vector<cl::Platform> platforms;
err = cl::Platform::get(&platforms);
cl::Platform platform = platforms[0]; // Choose the first platform
std::vector<cl::Device> devices;
err = platform.getDevices(CL_DEVICE_TYPE_GPU, &devices);
cl::Device device = devices[0]; // Choose the first GPU device
cl::Context context(device);
cl::CommandQueue queue(context, device);
// Create OpenCL buffers for input and output image data
cl::Buffer input_buffer(context, CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR, sizeof(uchar) * width * height * 3, input, &err);
return;
}
我尝试检查缓冲区大小是否大于主机设备的内存,但即使在限制范围内。
输入定义为无符号 char3d 数组
uchar*** array3D = new uchar** [height];
for (int i = 0; i < height; ++i) {
array3D[i] = new uchar* [width];
for (int j = 0; j < width; ++j) {
array3D[i][j] = new uchar[3]; // Assuming 3 channels (RGB)
}
}
这是你的错误
uchar*** array3D = new uchar** [height];
for (int i = 0; i < height; ++i) {
array3D[i] = new uchar* [width];
for (int j = 0; j < width; ++j) {
array3D[i][j] = new uchar[3]; // Assuming 3 channels (RGB)
}
}
必须是连续的内存缓冲区
uchar* array3D = new uchar[width * height * 3];