#include <ATen/AccumulateType.h>
#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 inline bool all_integral(std::initializer_list<std::reference_wrapper<at::Scalar>> l)
{
for (at::Scalar& s : l) {
if (!s.isIntegral(true)) {
return false;
}
}
return true;
}
static at::Tensor& arange_out_op_api(at::Scalar start, at::Scalar end, at::Scalar step, at::Tensor& result)
{
EXEC_NPU_CMD(aclnnArange, start, end, step, result);
return result;
}
at::Tensor arange(const at::Scalar& start, const at::Scalar& end, const at::Scalar& step,
c10::optional<at::ScalarType> dtype, c10::optional<at::Layout> layout,
c10::optional<at::Device> device, c10::optional<bool> pin_memory)
{
DO_COMPATIBILITY(aclnnArange, acl_op::arange(start, end, step, dtype, layout, device, pin_memory));
c10::TensorOptions option =
c10::TensorOptions().dtype(dtype).device(device).layout(layout).pinned_memory(pin_memory);
AT_DISPATCH_ALL_TYPES_AND2(
at::kHalf, at::kBFloat16, c10::typeMetaToScalarType(option.dtype()), "arange_npu_nn", [&]() {
using accscalar_type = at::acc_type<scalar_t, false>;
auto start_value = start.to<accscalar_type>();
auto end_value = end.to<accscalar_type>();
auto step_value = step.to<accscalar_type>();
TORCH_CHECK(step_value > 0 || step_value < 0, "step must be nonzero", OPS_ERROR(ErrCode::VALUE));
TORCH_CHECK(std::isfinite(static_cast<double>(start_value)) && std::isfinite(static_cast<double>(end_value)),
"unsupported range: ", static_cast<double>(start_value), " -> ", static_cast<double>(end_value), OPS_ERROR(ErrCode::VALUE));
TORCH_CHECK(((step_value > 0) && (end_value >= start_value)) ||
((step_value < 0) && (end_value <= start_value)),
"upper bound and larger bound inconsistent with step sign", OPS_ERROR(ErrCode::VALUE));
});
at::Scalar start_opt = start;
at::Scalar end_opt = end;
at::Scalar step_opt = step;
bool set_to_integral_dtype = !option.has_dtype() && all_integral({start_opt, end_opt, step_opt});
if (set_to_integral_dtype) {
option = option.dtype(at::ScalarType::Long);
}
auto output_size = op_infer::infersize_arange(start, end, step, c10::typeMetaToScalarType(option.dtype()));
at::Tensor result = npu_preparation::apply_tensor_without_format(output_size, option);
arange_out_op_api(start, end, step, result);
return result;
}
at::Tensor arange(const at::Scalar& start, const at::Scalar& end, c10::optional<at::ScalarType> dtype,
c10::optional<at::Layout> layout, c10::optional<at::Device> device,
c10::optional<bool> pin_memory)
{
DO_COMPATIBILITY(aclnnArange, acl_op::arange(start, end, dtype, layout, device, pin_memory));
return op_api::arange(start, end, 1, dtype, layout, device, pin_memory);
}
at::Tensor arange(const at::Scalar& end, c10::optional<at::ScalarType> dtype, c10::optional<at::Layout> layout,
c10::optional<at::Device> device, c10::optional<bool> pin_memory)
{
DO_COMPATIBILITY(aclnnArange, acl_op::arange(end, dtype, layout, device, pin_memory));
return op_api::arange(0, end, dtype, layout, device, pin_memory);
}
at::Tensor& arange_out(const at::Scalar& start, const at::Scalar& end, const at::Scalar& step, at::Tensor& out)
{
DO_COMPATIBILITY(aclnnArange, acl_op::arange_out(start, end, step, out));
AT_DISPATCH_ALL_TYPES_AND2(at::kHalf, at::kBFloat16, out.scalar_type(), "arange_out_npu_nn", [&]() {
using accscalar_type = at::acc_type<scalar_t, false>;
auto start_value = start.to<accscalar_type>();
auto end_value = end.to<accscalar_type>();
auto step_value = step.to<accscalar_type>();
TORCH_CHECK(step_value > 0 || step_value < 0, "step must be nonzero", OPS_ERROR(ErrCode::VALUE));
TORCH_CHECK(std::isfinite(static_cast<double>(start_value)) && std::isfinite(static_cast<double>(end_value)),
"unsupported range: ", static_cast<double>(start_value), " -> ", static_cast<double>(end_value), OPS_ERROR(ErrCode::VALUE));
TORCH_CHECK(((step_value > 0) && (end_value >= start_value)) ||
((step_value < 0) && (end_value <= start_value)),
"upper bound and larger bound inconsistent with step sign", OPS_ERROR(ErrCode::VALUE));
});
auto output_size = op_infer::infersize_arange(start, end, step, out.scalar_type());
int64_t size = output_size[0];
int64_t numel = out.numel();
if (numel != size) {
if (numel > 0) {
TORCH_WARN("The number of elements in the out tensor of shape ", out.sizes(),
" is ", numel, " which does not match the computed number of elements ", size,
". Note that this may occur as a result of rounding error. "
"The out tensor will be resized to a tensor of shape (", size, ",).");
}
out.resize_(output_size);
}
arange_out_op_api(start, end, step, out);
return out;
}
static at::Tensor& arange_start_end_out(at::Scalar start, at::Scalar end, at::Tensor& result)
{
at::Scalar step = 1;
return op_api::arange_out(start, end, step, result);
}
at::Tensor& arange_out(const at::Scalar& end, at::Tensor& out)
{
DO_COMPATIBILITY(aclnnArange, acl_op::arange_out(end, out));
return arange_start_end_out(0, end, out);
}
}