#include "op_plugin/AclOpsInterface.h"
#include "op_plugin/utils/OpAdapter.h"
namespace acl_op {
using npu_preparation = at_npu::native::OpPreparation;
using npu_utils = at_npu::native::NpuUtils;
namespace {
at::Tensor& linspace_npu_out_nocheck(
at::Tensor& result,
const at::Scalar& start,
const at::Scalar& end,
int64_t steps)
{
if (steps != 0) {
if (steps == 1) {
acl_op::fill_(result, start);
} else {
c10::SmallVector<int64_t, N> size_vec = {steps};
at_npu::native::OpCommand cmd;
cmd.Name("LinSpace")
.Input(start, at::ScalarType::Float)
.Input(end, at::ScalarType::Float)
.Input(size_vec, at::ScalarType::Int)
.Output(result)
.Run();
}
}
return result;
}
}
at::Tensor& linspace_out(const at::Scalar& start, const at::Scalar& end, int64_t steps, at::Tensor& out)
{
TORCH_CHECK(steps >= 0, "number of steps must be non-negative"
+ OPS_ERROR(ErrCode::VALUE));
if (out.numel() != steps) {
out.resize_({steps});
}
bool out_is_not_float = (out.dtype() != at::kFloat) ? true : false;
at::Tensor out_cast = out;
if (out_is_not_float) {
out_cast = at_npu::native::custom_ops::_npu_dtype_cast(out, at::kFloat);
}
if (!npu_utils::check_match(&out_cast)) {
at::Tensor contiguous_out = npu_utils::format_contiguous(out_cast);
linspace_npu_out_nocheck(contiguous_out, start, end, steps);
npu_utils::format_fresh_view(out_cast, contiguous_out);
} else {
linspace_npu_out_nocheck(out_cast, start, end, steps);
}
if (out_is_not_float) {
out_cast = at_npu::native::custom_ops::_npu_dtype_cast(out_cast, out.scalar_type());
out.copy_(out_cast);
} else {
out = out_cast;
}
return out;
}
at::Tensor linspace(
const at::Scalar& start,
const at::Scalar& end,
int64_t steps,
c10::optional<at::ScalarType> dtype,
c10::optional<at::Layout> layout,
c10::optional<at::Device> device,
c10::optional<bool> pin_memory)
{
TORCH_CHECK(steps >= 0, "number of steps must be non-negative"
+ OPS_ERROR(ErrCode::VALUE));
auto device_value = c10::device_or_default(device);
at::TensorOptions option = c10::TensorOptions()
.dtype(dtype).layout(layout).device(device_value).pinned_memory(pin_memory);
at::Tensor result = npu_preparation::apply_tensor_with_format({steps}, option, ACL_FORMAT_ND);
at::Tensor result_cast = result;
bool result_is_not_float = (result.dtype() != at::kFloat) ? true : false;
if (result_is_not_float) {
result_cast = at_npu::native::custom_ops::_npu_dtype_cast(result, at::kFloat);
}
linspace_npu_out_nocheck(result_cast, start, end, steps);
if (result_is_not_float) {
result_cast = at_npu::native::custom_ops::_npu_dtype_cast(result_cast, result.scalar_type());
}
result = result_cast;
return result;
}
}