vol2col.cuh

#pragma once

#include <THC/THCGeneral.h>
#include <THC/THCDeviceUtils.cuh>

#include <ATen/ATen.h>
#include <ATen/TensorUtils.h>
#include <ATen/Utils.h>

#include <ATen/cuda/CUDAContext.h>
#include <ATen/cuda/detail/KernelUtils.h>
#include <ATen/cuda/CUDAApplyUtils.cuh>
#include <ATen/cuda/detail/IndexUtils.cuh>
#include <ATen/cuda/detail/TensorInfo.cuh>

#include <c10/macros/Macros.h>

namespace at {
namespace native {

using namespace at::cuda::detail;

// Kernel for fast unfold+copy on volumes
template <typename T>
__global__ void vol2col_kernel(
    const int n,
    const T* data_vol,
    const int depth,
    const int height,
    const int width,
    const int ksize_t,
    const int ksize_h,
    const int ksize_w,
    const int pad_t,
    const int pad_h,
    const int pad_w,
    const int stride_t,
    const int stride_h,
    const int stride_w,
    const int dilation_t,
    const int dilation_h,
    const int dilation_w,
    const int depth_col,
    const int height_col,
    const int width_col,
    T* data_col) {
  CUDA_KERNEL_LOOP(index, n) {
    int w_out = index % width_col;
    index /= width_col;
    int h_out = index % height_col;
    index /= height_col;
    int t_out = index % depth_col;
    int channel_in = index / depth_col;
    int channel_out = channel_in * ksize_t * ksize_h * ksize_w;
    int t_in = t_out * stride_t - pad_t;
    int h_in = h_out * stride_h - pad_h;
    int w_in = w_out * stride_w - pad_w;
    data_col +=
        ((channel_out * depth_col + t_out) * height_col + h_out) * width_col +
        w_out;
    data_vol += ((channel_in * depth + t_in) * height + h_in) * width + w_in;
    for (int i = 0; i < ksize_t; ++i) {
      for (int j = 0; j < ksize_h; ++j) {
        for (int k = 0; k < ksize_w; ++k) {
          int t = t_in + i * dilation_t;
          int h = h_in + j * dilation_h;
          int w = w_in + k * dilation_w;
          *data_col = (t >= 0 && h >= 0 && w >= 0 && t < depth && h < height &&
                       w < width)
              ? data_vol
                    [i * dilation_t * height * width + j * dilation_h * width +
                     k * dilation_w]
              : static_cast<T>(0);
          data_col += depth_col * height_col * width_col;
        }
      }
    }
  }
}

template <typename T>
void vol2col(
    cudaStream_t stream,
    const T* data_vol,
    const int channels,
    const int depth,
    const int height,
    const int width,
    const int depth_col,
    const int height_col,
    const int width_col,
    const int ksize_t,
    const int ksize_h,
    const int ksize_w,
    const int pad_t,
    const int pad_h,
    const int pad_w,
    const int stride_t,
    const int stride_h,
    const int stride_w,
    const int dilation_t,
    const int dilation_h,
    const int dilation_w,
    T* data_col) {
  // We are going to launch channels * depth_col * height_col * width_col
  // kernels, each kernel responsible for copying a single-channel grid.
  int num_kernels = channels * depth_col * height_col * width_col;
  // Launch
  vol2col_kernel<<<GET_BLOCKS(num_kernels), CUDA_NUM_THREADS, 0, stream>>>(
      num_kernels,
      data_vol,
      depth,
      height,
      width,
      ksize_t,
      ksize_h,
      ksize_w,
      pad_t,
      pad_h,
      pad_w,
      stride_t,
      stride_h,
      stride_w,
      dilation_t,
      dilation_h,
      dilation_w,
      depth_col,
      height_col,
      width_col,
      data_col);
  AT_CUDA_CHECK(cudaGetLastError());
}

template <typename T, typename accT>
__global__ void vol2im_kernel(
    const unsigned n,
    const T* data_col,
    const unsigned depth,
    const unsigned height,
    const unsigned width,
    const unsigned channels,
    const unsigned kernel_t,
    const unsigned kernel_h,
    const unsigned kernel_w,
    const unsigned pad_t,
    const unsigned pad_h,
    const unsigned pad_w,
    const unsigned stride_t,
    const unsigned stride_h,
    const unsigned stride_w,
    const unsigned dilation_t,
    const unsigned dilation_h,
    const unsigned dilation_w,
    const unsigned depth_col,
    const unsigned height_col,
    const unsigned width_col,
    T* data_vol) {
  CUDA_KERNEL_LOOP(index, n) {
    accT val = static_cast<accT>(0);
    const unsigned w_im = index % width + pad_w;
    const unsigned h_im = (index / width) % height + pad_h;
    const unsigned t_im = (index / width / height) % depth + pad_t;
    const unsigned c_im = index / (width * height * depth);
    unsigned kernel_extent_w = (kernel_w - 1) * dilation_w + 1;
    unsigned kernel_extent_h = (kernel_h - 1) * dilation_h + 1;
    unsigned kernel_extent_t = (kernel_t - 1) * dilation_t + 1;
    // compute the start and end of the output
    const unsigned w_col_start =
        (w_im < kernel_extent_w) ? 0 : (w_im - kernel_extent_w) / stride_w + 1;
    const unsigned w_col_end = std::min(w_im / stride_w + 1, width_col);
    const unsigned h_col_start =
        (h_im < kernel_extent_h) ? 0 : (h_im - kernel_extent_h) / stride_h + 1;
    const unsigned h_col_end = std::min(h_im / stride_h + 1, height_col);
    const unsigned t_col_start =
        (t_im < kernel_extent_t) ? 0 : (t_im - kernel_extent_t) / stride_t + 1;
    const unsigned t_col_end = std::min(t_im / stride_t + 1, depth_col);
    // TODO: use LCM of stride and dilation to avoid unnecessary loops
    for (unsigned t_col = t_col_start; t_col < t_col_end; t_col += 1) {
      for (unsigned h_col = h_col_start; h_col < h_col_end; h_col += 1) {
        for (unsigned w_col = w_col_start; w_col < w_col_end; w_col += 1) {
          unsigned t_k = (t_im - t_col * stride_t);
          unsigned h_k = (h_im - h_col * stride_h);
          unsigned w_k = (w_im - w_col * stride_w);
          if (t_k % dilation_t == 0 && h_k % dilation_h == 0 &&
              w_k % dilation_w == 0) {
            t_k /= dilation_t;
            h_k /= dilation_h;
            w_k /= dilation_w;
            const int64_t idx_k =
                ((c_im * kernel_t + t_k) * kernel_h + h_k) * kernel_w + w_k;
            const int64_t data_col_index =
                ((idx_k * depth_col + t_col) *
                    height_col + h_col) *
                  width_col + w_col;
            val += data_col[data_col_index];
          }
        }
      }
    }
    data_vol[index] = static_cast<T>(val);
  }
}

template <typename T, typename accT>
void col2vol(
    cudaStream_t stream,
    const T* data_col,
    const int64_t channels,
    const int64_t depth,
    const int64_t height,
    const int64_t width,
    const int64_t output_depth,
    const int64_t output_height,
    const int64_t output_width,
    const int64_t patch_t,
    const int64_t patch_h,
    const int64_t patch_w,
    const int64_t pad_t,
    const int64_t pad_h,
    const int64_t pad_w,
    const int64_t stride_t,
    const int64_t stride_h,
    const int64_t stride_w,
    const int64_t dilation_t,
    const int64_t dilation_h,
    const int64_t dilation_w,
    T* data_vol) {
  const auto num_kernels = channels * depth * height * width;

  auto check_fits_in_unsigned =
    [](int64_t val, const char * name) {
      constexpr auto umax = std::numeric_limits<unsigned>::max();
      TORCH_CHECK(val >= 0 && val <= umax,
                  name, " must fit in a 32-bit unsigned value");
    };
  check_fits_in_unsigned(num_kernels, "input size");
  check_fits_in_unsigned(
      channels * patch_t * patch_h * patch_w, "channels x kernel size");

  // To avoid involving atomic operations, we will launch one kernel per
  // bottom dimension, and then in the kernel add up the top dimensions.
  vol2im_kernel<T, accT>
      <<<GET_BLOCKS(num_kernels), CUDA_NUM_THREADS, 0, stream>>>(
          num_kernels,
          data_col,
          depth,
          height,
          width,
          channels,
          patch_t,
          patch_h,
          patch_w,
          pad_t,
          pad_h,
          pad_w,
          stride_t,
          stride_h,
          stride_w,
          dilation_t,
          dilation_h,
          dilation_w,
          output_depth,
          output_height,
          output_width,
          data_vol);
  AT_CUDA_CHECK(cudaGetLastError());
}

} // namespace native
} // namespace at