#include "csrc/OpApiCommon.h"
#include "csrc/functions.h"
namespace {
constexpr int8_t INPUT_DIM = 4;
constexpr int16_t CHANNEL_256 = 256;
constexpr int16_t CHANNEL_512 = 512;
}
std::tuple<at::Tensor, at::Tensor> modulated_deformable_conv2d(const at::Tensor& input, const at::Tensor& offset,
const at::Tensor& mask, const at::Tensor& weight, const c10::optional<at::Tensor>& bias_opt,
at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation,
int64_t groups, int64_t deformable_groups, int64_t with_bias)
{
TORCH_CHECK_NPU(input);
TORCH_CHECK_NPU(offset);
TORCH_CHECK_NPU(mask);
TORCH_CHECK_NPU(weight);
TORCH_CHECK(input.dim() == INPUT_DIM, "input must to be a 4D Tensor, but got: ", input.dim());
TORCH_CHECK(offset.dim() == INPUT_DIM, "offset has to be a 4D Tensor, but got: ", offset.dim());
TORCH_CHECK(mask.dim() == INPUT_DIM, "mask has to be a 4D Tensor, but got: ", mask.dim());
TORCH_CHECK(weight.dim() == INPUT_DIM, "weight has to be a 4D Tensor, but got: ", weight.dim());
TORCH_CHECK(stride[0] > 0 && stride[1] > 0, "stride must be greater than 0");
TORCH_CHECK(kernel_size[0] > 0 && kernel_size[1] > 0, "kernel_size must be greater than 0");
TORCH_CHECK(dilation[0] > 0 && dilation[1] > 0, "dilation must be greater than 0");
const at::Tensor& bias = bias_opt.value_or(at::Tensor());
uint32_t n = static_cast<uint32_t>(input.size(0));
uint32_t c_in = static_cast<uint32_t>(input.size(3));
uint32_t h_in = static_cast<uint32_t>(input.size(1));
uint32_t w_in = static_cast<uint32_t>(input.size(2));
uint32_t h_out = static_cast<uint32_t>(offset.size(1));
uint32_t w_out = static_cast<uint32_t>(offset.size(2));
uint32_t c_out = static_cast<uint32_t>(weight.size(0));
uint32_t kh = static_cast<uint32_t>(weight.size(1));
uint32_t kw = static_cast<uint32_t>(weight.size(2));
TORCH_CHECK(kh == kernel_size[0] && kw == kernel_size[1], "kernel size mismatch");
TORCH_CHECK(mask.size(-1) == kh * kw, "The shape of the mask is invalid");
TORCH_CHECK(groups > 0, "groups must be greater than 0");
TORCH_CHECK(c_out % groups == 0, "weight's out channel should be divided by groups");
TORCH_CHECK(c_in % groups == 0, "input's channel should be divided by groups");
bool modulated = true;
if ((groups == 1) && ((c_in == CHANNEL_256) || (c_in == CHANNEL_512))) {
at::Tensor output = at::empty({n, h_out, w_out, c_out}, input.options());
at::Tensor offset_output = at::empty({n, h_out * w_out, kh * kw, c_in}, input.options());
EXEC_NPU_CMD(aclnnDeformableConv2dV2, input, offset, mask, weight, bias, kernel_size, stride, padding, dilation,
groups, deformable_groups, modulated, with_bias, output, offset_output);
output = output.permute({0, 3, 1, 2});
return std::tie(output, offset_output);
} else {
at::Tensor output = at::empty({n, h_out, c_out, w_out}, input.options());
at::Tensor offset_output = at::empty({n, h_out, w_out, groups, kh * kw * c_in / groups}, input.options());
EXEC_NPU_CMD(aclnnDeformableConv2d, input, weight, bias, offset, mask, kernel_size, stride, padding, dilation,
groups, deformable_groups, modulated, with_bias, output, offset_output);
output = output.permute({0, 2, 1, 3});
return std::tie(output, offset_output);
}
}
std::tuple<at::Tensor, at::Tensor, at::Tensor, at::Tensor, at::Tensor> modulated_deformable_conv2d_backward(
const at::Tensor& input, const at::Tensor& offset, const at::Tensor& mask, const at::Tensor& weight,
const c10::optional<at::Tensor>& bias_opt, const at::Tensor& offset_output, const at::Tensor& grad_y,
at::IntArrayRef kernel_size, at::IntArrayRef stride, at::IntArrayRef padding, at::IntArrayRef dilation,
int64_t groups, int64_t deformable_groups, int64_t with_bias)
{
TORCH_CHECK_NPU(input);
TORCH_CHECK_NPU(offset);
TORCH_CHECK_NPU(weight);
TORCH_CHECK(input.dim() == INPUT_DIM, "input must to be a 4D Tensor, but got: ", input.dim());
TORCH_CHECK(offset.dim() == INPUT_DIM, "offset has to be a 4D Tensor, but got: ", offset.dim());
TORCH_CHECK(mask.dim() == INPUT_DIM, "mask has to be a 4D Tensor, but got: ", mask.dim());
TORCH_CHECK(weight.dim() == INPUT_DIM, "weight has to be a 4D Tensor, but got: ", weight.dim());
TORCH_CHECK(stride[0] > 0 && stride[1] > 0, "stride must be greater than 0");
TORCH_CHECK(kernel_size[0] > 0 && kernel_size[1] > 0, "kernel_size must be greater than 0");
TORCH_CHECK(dilation[0] > 0 && dilation[1] > 0, "dilation must be greater than 0");
const at::Tensor& bias = bias_opt.value_or(at::Tensor());
at::Tensor grad_bias = at::Tensor();
auto input_sizes = input.sizes();
auto offset_sizes = offset.sizes();
auto weight_sizes = weight.sizes();
auto mask_sizes = mask.sizes();
at::Tensor grad_input = at::zeros(input_sizes, input.options());
at::Tensor grad_offset = at::empty(
{offset_sizes[0], offset_sizes[1], offset_sizes[2], 2, kernel_size[0], kernel_size[1]}, offset.options());
at::Tensor grad_weight = at::zeros(weight_sizes, weight.options());
at::Tensor grad_mask = at::zeros(mask_sizes, mask.options());
bool modulated = true;
uint32_t c_in = static_cast<uint32_t>(input.size(3));
if ((groups == 1) && ((c_in == CHANNEL_256) || (c_in == CHANNEL_512))) {
EXEC_NPU_CMD(aclnnDeformableConv2dGradV2, input, weight, bias, offset, mask, grad_y, kernel_size, stride,
padding, dilation, groups, deformable_groups, modulated, with_bias, grad_input, grad_weight, grad_bias,
grad_offset, grad_mask);
} else {
EXEC_NPU_CMD(aclnnDeformableConv2dGrad, input, weight, bias, offset, mask, offset_output, grad_y, kernel_size,
stride, padding, dilation, groups, deformable_groups, modulated, with_bias, grad_input, grad_weight,
grad_bias, grad_offset, grad_mask);
}
grad_input = grad_input.permute({0, 3, 1, 2});
grad_weight = grad_weight.permute({0, 3, 1, 2});
grad_offset = grad_offset.permute({0, 4, 5, 3, 1, 2});
grad_mask = grad_mask.permute({0, 3, 1, 2});
return std::tie(grad_input, grad_weight, grad_bias, grad_offset, grad_mask);
}