#include "op_plugin/AclOpsInterface.h"
#include "op_plugin/OpApiInterface.h"
#include "op_plugin/utils/op_api_common.h"
namespace op_api {
using npu_preparation = at_npu::native::OpPreparation;
static const int MODE_TRUNC = 1;
static const int MODE_FLOOR = 2;
static void check_rounding_mode_npu(c10::optional<c10::string_view> rounding_mode)
{
TORCH_CHECK((!rounding_mode.has_value() || *rounding_mode == "trunc" || *rounding_mode == "floor"),
"div expected rounding_mode to be one of None, 'trunc', or 'floor' "
"but found '",
*rounding_mode, "'", OPS_ERROR(ErrCode::PARAM));
}
static at::Tensor& div_out_npu_opapi_nocheck(const at::Tensor& self, const at::Tensor& other, at::Tensor& result) {
if (other.dim() == 0 && !torch_npu::utils::is_npu(other)) {
c10::Scalar others = other.item();
EXEC_NPU_CMD(aclnnDivs, self, others, result);
} else {
EXEC_NPU_CMD(aclnnDiv, self, other, result);
}
return result;
}
static at::Tensor self_tensor_to_device(const at::Tensor& tensor, const at::ScalarType result_type,
const c10::Device device)
{
if (npu_preparation::is_scalar_wrapped_to_tensor(tensor)) {
at::Scalar scalar = tensor.item();
return npu_preparation::copy_scalar_to_device(scalar, result_type, device);
}
return tensor;
}
at::Tensor& div_out(const at::Tensor& self, const at::Tensor& other, at::Tensor& result) {
DO_COMPATIBILITY(aclnnDivs, acl_op::div_out(self, other, result));
DO_COMPATIBILITY(aclnnDiv, acl_op::div_out(self, other, result));
std::vector<at::Tensor> tensor_list = {self, other};
auto maybe_names = op_plugin::utils::compute_names_npu(tensor_list);
auto output_size = op_infer::broadcast_ops_npu_output_size(self, other);
at::ScalarType result_type = at::native::result_type(self, other);
if (!isFloatingType(result_type) && !isComplexType(result_type)) {
result_type = at::ScalarType::Float;
}
if (isFloatingType(result.scalar_type()) || isComplexType(result.scalar_type())) {
result_type = result.scalar_type();
}
at::Tensor self_cp = self_tensor_to_device(self, result_type, result.device());
npu_preparation::check_tensor({self}, result, result_type, output_size);
div_out_npu_opapi_nocheck(self_cp, other, result);
at::namedinference::propagate_names_if_nonempty(result, maybe_names);
return result;
}
at::Tensor& div_out(const at::Tensor& self, const at::Tensor& other, c10::optional<c10::string_view> rounding_mode,
at::Tensor& result)
{
DO_COMPATIBILITY(aclnnDivMods, acl_op::div_out(self, other, rounding_mode, result));
DO_COMPATIBILITY(aclnnDivMod, acl_op::div_out(self, other, rounding_mode, result));
std::vector<at::Tensor> tensor_list = {self, other};
auto maybe_names = op_plugin::utils::compute_names_npu(tensor_list);
if (rounding_mode.has_value() && *rounding_mode != "floor" && *rounding_mode != "trunc") {
TORCH_CHECK(false,
"div expected rounding_mode to be one of None, 'trunc', or 'floor' "
"but found '",
*rounding_mode, "'", OPS_ERROR(ErrCode::PARAM));
}
auto outputSize = op_infer::broadcast_ops_npu_output_size(self, other);
at::ScalarType result_type = at::native::result_type(self, other);
at::Tensor self_cp = self_tensor_to_device(self, result_type, result.device());
npu_preparation::check_tensor({self}, result, result.scalar_type(), outputSize);
int mode = 0;
if (rounding_mode.has_value() && *rounding_mode == "floor") {
mode = MODE_FLOOR;
} else if (rounding_mode.has_value() && *rounding_mode == "trunc") {
mode = MODE_TRUNC;
}
if (other.dim() == 0 && !torch_npu::utils::is_npu(other)) {
c10::Scalar others = other.item();
EXEC_NPU_CMD(aclnnDivMods, self_cp, others, mode, result);
} else {
EXEC_NPU_CMD(aclnnDivMod, self_cp, other, mode, result);
}
at::namedinference::propagate_names_if_nonempty(result, maybe_names);
return result;
}
at::Tensor div(const at::Tensor& self, const at::Tensor& other) {
DO_COMPATIBILITY(aclnnDivs, acl_op::div(self, other));
DO_COMPATIBILITY(aclnnDiv, acl_op::div(self, other));
std::vector<at::Tensor> tensor_list = {self, other};
auto maybe_names = op_plugin::utils::compute_names_npu(tensor_list);
bool isSelfWrapped = npu_preparation::is_scalar_wrapped_to_tensor(self);
at::Tensor outputTensor = isSelfWrapped ? other : self;
auto outputSize = op_infer::broadcast_ops_npu_output_size(self, other);
at::ScalarType high_type = at::native::result_type(self, other);
if (!isFloatingType(high_type) && !isComplexType(high_type)) {
high_type = at::ScalarType::Float;
}
at::Tensor self_cp = self_tensor_to_device(self, high_type, outputTensor.device());
at::Tensor result = npu_preparation::apply_tensor_without_format(outputSize, outputTensor.options().dtype(high_type));
div_out_npu_opapi_nocheck(self_cp, other, result);
at::namedinference::propagate_names_if_nonempty(result, maybe_names);
return result;
}
at::Tensor div(const at::Tensor& self, const at::Tensor& other, c10::optional<c10::string_view> rounding_mode)
{
DO_COMPATIBILITY(aclnnDivMods, acl_op::div(self, other, rounding_mode));
DO_COMPATIBILITY(aclnnDivMod, acl_op::div(self, other, rounding_mode));
std::vector<at::Tensor> tensor_list = {self, other};
auto maybe_names = op_plugin::utils::compute_names_npu(tensor_list);
if (rounding_mode.has_value() && *rounding_mode != "floor" && *rounding_mode != "trunc") {
TORCH_CHECK(false,
"div expected rounding_mode to be one of None, 'trunc', or 'floor' "
"but found '",
*rounding_mode, "'", OPS_ERROR(ErrCode::PARAM));
}
bool isSelfWrapped = npu_preparation::is_scalar_wrapped_to_tensor(self);
at::Tensor outputTensor = isSelfWrapped ? other : self;
auto outputSize = op_infer::broadcast_ops_npu_output_size(self, other);
at::ScalarType high_type = at::native::result_type(self, other);
at::Tensor self_cp = self_tensor_to_device(self, high_type, outputTensor.device());
int mode = 0;
if (rounding_mode.has_value() && *rounding_mode == "floor") {
mode = MODE_FLOOR;
} else if (rounding_mode.has_value() && *rounding_mode == "trunc") {
mode = MODE_TRUNC;
} else {
if (!isFloatingType(high_type) && !isComplexType(high_type)) {
high_type = at::ScalarType::Float;
}
}
at::Tensor result = npu_preparation::apply_tensor_without_format(outputSize, outputTensor.options().dtype(high_type));
if (other.dim() == 0 && !torch_npu::utils::is_npu(other)) {
c10::Scalar others = other.item();
EXEC_NPU_CMD(aclnnDivMods, self_cp, others, mode, result);
} else {
EXEC_NPU_CMD(aclnnDivMod, self_cp, other, mode, result);
}
at::namedinference::propagate_names_if_nonempty(result, maybe_names);
return result;
}
static at::Tensor& inplace_div_out_npu_no_check(at::Tensor& self, const at::Tensor& other) {
if (other.dim() == 0 && !torch_npu::utils::is_npu(other)) {
c10::Scalar others = other.item();
EXEC_NPU_CMD(aclnnInplaceDivs, self, others);
} else {
EXEC_NPU_CMD(aclnnInplaceDiv, self, other);
}
return self;
}
static at::Tensor& inplace_div_out_mode_npu_no_check(at::Tensor& self, const at::Tensor& other, int mode) {
if (other.dim() == 0 && !torch_npu::utils::is_npu(other)) {
c10::Scalar others = other.item();
EXEC_NPU_CMD(aclnnInplaceDivMods, self, others, mode);
} else {
EXEC_NPU_CMD(aclnnInplaceDivMod, self, other, mode);
}
return self;
}
at::Tensor& div_(at::Tensor& self, const at::Tensor& other) {
DO_COMPATIBILITY(aclnnInplaceDivs, acl_op::div_(self, other));
DO_COMPATIBILITY(aclnnInplaceDiv, acl_op::div_(self, other));
std::vector<at::Tensor> tensor_list = {self, other};
auto maybe_names = op_plugin::utils::compute_names_npu(tensor_list);
const std::initializer_list<at::Tensor> inputs = {self, other};
const std::initializer_list<at::Tensor> outputs = {self};
npu_preparation::check_memory(inputs, outputs);
inplace_div_out_npu_no_check(self, other);
at::namedinference::propagate_names_if_nonempty(self, maybe_names);
return self;
}
at::Tensor& div_(at::Tensor& self, const at::Tensor& other, c10::optional<c10::string_view> rounding_mode) {
DO_COMPATIBILITY(aclnnInplaceDivMods, acl_op::div_(self, other, rounding_mode));
DO_COMPATIBILITY(aclnnInplaceDivMod, acl_op::div_(self, other, rounding_mode));
std::vector<at::Tensor> tensor_list = {self, other};
auto maybe_names = op_plugin::utils::compute_names_npu(tensor_list);
check_rounding_mode_npu(rounding_mode);
const std::initializer_list<at::Tensor> inputs = {self, other};
const std::initializer_list<at::Tensor> outputs = {self};
npu_preparation::check_memory(inputs, outputs);
int mode = 0;
if (rounding_mode.has_value() && *rounding_mode == "floor") {
mode = MODE_FLOOR;
} else if (rounding_mode.has_value() && *rounding_mode == "trunc") {
mode = MODE_TRUNC;
}
inplace_div_out_mode_npu_no_check(self, other, mode);
at::namedinference::propagate_names_if_nonempty(self, maybe_names);
return self;
}
at::Tensor div(const at::Tensor& self, const at::Scalar& other) {
DO_COMPATIBILITY(aclnnDivs, acl_op::div(self, other));
auto outputSize = op_infer::input_same_output_size(self);
at::ScalarType high_type = at::native::result_type(self, other);
if (!isFloatingType(high_type) && !isComplexType(high_type)) {
high_type = at::ScalarType::Float;
}
at::Tensor result = npu_preparation::apply_tensor_without_format(outputSize, self.options().dtype(high_type));
EXEC_NPU_CMD(aclnnDivs, self, other, result);
at::namedinference::propagate_names(result, self);
return result;
}
at::Tensor div(const at::Tensor& self, const at::Scalar& other, c10::optional<c10::string_view> rounding_mode) {
DO_COMPATIBILITY(aclnnDivMods, acl_op::div(self, other, rounding_mode));
check_rounding_mode_npu(rounding_mode);
auto outputSize = op_infer::input_same_output_size(self);
at::ScalarType high_type = at::native::result_type(self, other);
int mode = 0;
if (rounding_mode.has_value() && *rounding_mode == "floor") {
mode = MODE_FLOOR;
} else if (rounding_mode.has_value() && *rounding_mode == "trunc") {
mode = MODE_TRUNC;
} else {
if (!isFloatingType(high_type) && !isComplexType(high_type)) {
high_type = at::ScalarType::Float;
}
}
at::Tensor result = npu_preparation::apply_tensor_without_format(outputSize, self.options().dtype(high_type));
EXEC_NPU_CMD(aclnnDivMods, self, other, mode, result);
at::namedinference::propagate_names(result, self);
return result;
}
at::Tensor& div_(at::Tensor& self, const at::Scalar& other) {
DO_COMPATIBILITY(aclnnInplaceDivs, acl_op::div_(self, other));
EXEC_NPU_CMD(aclnnInplaceDivs, self, other);
return self;
}
at::Tensor& div_(at::Tensor& self, const at::Scalar& other, c10::optional<c10::string_view> rounding_mode) {
DO_COMPATIBILITY(aclnnInplaceDivMods, acl_op::div_(self, other, rounding_mode));
check_rounding_mode_npu(rounding_mode);
int mode = 0;
if (rounding_mode.has_value() && *rounding_mode == "floor") {
mode = MODE_FLOOR;
} else if (rounding_mode.has_value() && *rounding_mode == "trunc") {
mode = MODE_TRUNC;
}
EXEC_NPU_CMD(aclnnInplaceDivMods, self, other, mode);
return self;
}
}