* Copyright (c) 2026 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/
* \file add.cpp
* \brief
*/
#include <ATen/Operators.h>
#include <torch/all.h>
#include <torch/library.h>
#include "torch_npu/csrc/core/npu/NPUStream.h"
#include "torch_npu/csrc/framework/OpCommand.h"
#include "kernel_operator.h"
#include "platform/platform_ascendc.h"
#include <type_traits>
#include "atvoss.h"
TORCH_LIBRARY_FRAGMENT(EXTENSION_MODULE_NAME, m)
{
m.def("abs(Tensor x) -> Tensor");
}
torch::Tensor abs_meta(const torch::Tensor& x)
{
auto y = torch::empty_like(x);
return y;
}
TORCH_LIBRARY_IMPL(EXTENSION_MODULE_NAME, Meta, m)
{
m.impl("abs", abs_meta);
}
static constexpr int32_t TILE_SIZE = 32;
static constexpr int32_t SHAPE_SIZE = 2;
template <typename T>
struct AbsConfig {
using Dtype = T;
using TileShape = Atvoss::Shape<TILE_SIZE>;
struct AbsCompute {
template <template <typename> class Tensor>
__host_aicore__ constexpr auto Compute() const
{
auto in1 = Atvoss::PlaceHolder<1, Tensor<Dtype>, Atvoss::ParamUsage::IN>();
auto out1 = Atvoss::PlaceHolder<2, Tensor<Dtype>, Atvoss::ParamUsage::OUT>();
return (out1 = Abs(in1));
};
};
static constexpr Atvoss::Ele::DefaultBlockPolicy<TileShape> blockPolicy{TileShape{}};
static constexpr Atvoss::Ele::DefaultKernelPolicy kernelPolicy{Atvoss::Ele::DefaultSegmentPolicy::UniformSegment};
using ArchTag = Atvoss::Arch::DAV_3510;
using BlockOp = Atvoss::Ele::BlockBuilder<AbsCompute, ArchTag, blockPolicy, Atvoss::Ele::DefaultBlockConfig>;
using KernelOp = Atvoss::Ele::KernelBuilder<BlockOp, kernelPolicy>;
using DeviceOp = Atvoss::DeviceAdapter<KernelOp>;
};
torch::Tensor abs_npu(const torch::Tensor& x)
{
auto y = abs_meta(x);
uint64_t shape[SHAPE_SIZE] = {};
std::copy(x.sizes().begin(), x.sizes().end(), shape);
Atvoss::Tensor<float> t1(x.data_ptr<float>(), shape, SHAPE_SIZE);
Atvoss::Tensor<float> t2(y.data_ptr<float>(), shape, SHAPE_SIZE);
auto arguments = Atvoss::ArgumentsBuilder{}.inputOutput(t1, t2).build();
using Config = AbsConfig<float>;
auto stream = c10_npu::getCurrentNPUStream().stream(false);
Config::DeviceOp deviceOp;
deviceOp.Run(arguments, stream);
return y;
}
TORCH_LIBRARY_IMPL(EXTENSION_MODULE_NAME, PrivateUse1, m)
{
m.impl("abs", abs_npu);
}