Method configure

src/gpu/cl/kernels/ClDepthConcatenateKernel.cpp:71–116

}

void ClDepthConcatenateKernel::configure(const CLCompileContext &compile_context,
                                         ITensorInfo            *src,
                                         unsigned int            depth_offset,
                                         ITensorInfo            *dst)
{
    ARM_COMPUTE_ERROR_ON_NULLPTR(src, dst);
    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(src, depth_offset, dst));

    auto padding_info = get_padding_info({src, dst});

    _depth_offset = depth_offset;

    const unsigned int num_elems_processed_per_iteration = adjust_vec_size(16 / src->element_size(), src->dimension(0));

    // Add build options
    CLBuildOptions build_opts;
    build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(src->data_type()));
    build_opts.add_option("-DVEC_SIZE=" + support::cpp11::to_string(num_elems_processed_per_iteration));
    build_opts.add_option("-DVEC_SIZE_LEFTOVER=" +
                          support::cpp11::to_string(src->dimension(0) % num_elems_processed_per_iteration));
    if (is_data_type_quantized_asymmetric(src->data_type()) && src->quantization_info() != dst->quantization_info())
    {
        const UniformQuantizationInfo iq_info = src->quantization_info().uniform();
        const UniformQuantizationInfo oq_info = dst->quantization_info().uniform();

        build_opts.add_option("-DOFFSET_IN1=" + float_to_string_with_full_precision(iq_info.offset));
        build_opts.add_option("-DOFFSET_OUT=" + float_to_string_with_full_precision(oq_info.offset));
        build_opts.add_option("-DSCALE_IN1=" + float_to_string_with_full_precision(iq_info.scale));
        build_opts.add_option("-DSCALE_OUT=" + float_to_string_with_full_precision(oq_info.scale));
    }

    std::string kernel_name = "concatenate";

    // A macro guard to compile ONLY the kernel of interest
    build_opts.add_option("-D" + upper_string(kernel_name));

    // Create kernel
    _kernel = create_kernel(compile_context, kernel_name, build_opts.options());

    // Configure kernel window
    auto win = calculate_max_window(*dst, Steps(num_elems_processed_per_iteration));
    win.set(Window::DimZ, Window::Dimension(0, src->tensor_shape().z(), 1));
    ICLKernel::configure_internal(win);

    ARM_COMPUTE_ERROR_ON(has_padding_changed(padding_info));
}

Status ClDepthConcatenateKernel::validate(const arm_compute::ITensorInfo *src,
                                          unsigned int                    depth_offset,