111 lines
3.8 KiB
C++
111 lines
3.8 KiB
C++
// Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
|
|
//
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
// you may not use this file except in compliance with the License.
|
|
// You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
|
|
#include "paddle/phi/kernels/slice_kernel.h"
|
|
|
|
#include "glog/logging.h"
|
|
|
|
#include "paddle/common/flags.h"
|
|
#include "paddle/phi/backends/all_context.h"
|
|
#include "paddle/phi/core/kernel_registry.h"
|
|
#include "paddle/phi/kernels/funcs/slice_utils.h"
|
|
|
|
COMMON_DECLARE_bool(use_stride_kernel);
|
|
|
|
namespace phi {
|
|
|
|
template <typename Context>
|
|
void SliceStridedKernel(const Context& dev_ctx,
|
|
const DenseTensor& input,
|
|
const std::vector<int64_t>& axes,
|
|
const IntArray& starts_arr,
|
|
const IntArray& ends_arr,
|
|
const std::vector<int64_t>& infer_flags,
|
|
const std::vector<int64_t>& decrease_axis,
|
|
DenseTensor* out) {
|
|
if (!FLAGS_use_stride_kernel) {
|
|
PADDLE_THROW(common::errors::Fatal(
|
|
"FLAGS_use_stride_kernel is closed. Strided kernel "
|
|
"be called, something wrong has happened!"));
|
|
}
|
|
std::vector<int64_t> starts = starts_arr.GetData();
|
|
std::vector<int64_t> ends = ends_arr.GetData();
|
|
const auto& in_dims = input.dims();
|
|
|
|
auto new_axes = axes;
|
|
for (auto& item : new_axes) {
|
|
if (item < 0) {
|
|
item = std::max(int64_t(0), item + int64_t(in_dims.size()));
|
|
}
|
|
}
|
|
// axis = 0, dim_value = 3, st[0]=0, ed[0]=4
|
|
// The step seems to be regarded as 1 here
|
|
funcs::CheckAndUpdateSliceAttrs<int64_t>(
|
|
in_dims, new_axes, &starts, &ends, nullptr, nullptr);
|
|
|
|
std::vector<int64_t> output_dims = vectorize<int64_t>(input.dims());
|
|
std::vector<int64_t> output_stride = vectorize<int64_t>(input.strides());
|
|
int64_t output_offset = static_cast<int64_t>(input.offset());
|
|
|
|
for (size_t i = 0; i < new_axes.size(); ++i) {
|
|
output_offset = static_cast<int64_t>(
|
|
output_offset +
|
|
starts[i] * output_stride[new_axes[i]] * SizeOf(out->dtype()));
|
|
output_dims[new_axes[i]] = std::abs(ends[i] - starts[i]);
|
|
}
|
|
|
|
std::vector<uint8_t> decrease_flag(output_dims.size(), 0);
|
|
if (!decrease_axis.empty()) {
|
|
for (auto axis : decrease_axis) {
|
|
decrease_flag[axis] = 1;
|
|
}
|
|
|
|
std::vector<int64_t> new_shape;
|
|
std::vector<int64_t> new_stride;
|
|
for (size_t i = 0; i < output_dims.size(); ++i) {
|
|
if (decrease_flag[i] == 0) {
|
|
new_shape.push_back(output_dims[i]);
|
|
new_stride.push_back(output_stride[i]);
|
|
}
|
|
}
|
|
output_dims = new_shape;
|
|
output_stride = new_stride;
|
|
}
|
|
|
|
auto meta = out->meta();
|
|
meta.offset = output_offset;
|
|
auto tmp_dim = DDim(output_dims.data(), static_cast<int>(output_dims.size()));
|
|
// if (product(meta.dims) > 0 && meta.dims != tmp_dim) {
|
|
// PADDLE_THROW(
|
|
// common::errors::Fatal("Slice kernel stride compute diff, infer shape
|
|
// is
|
|
// "
|
|
// "%s, but compute is %s.",
|
|
// meta.dims,
|
|
// tmp_dim));
|
|
// }
|
|
meta.dims = tmp_dim;
|
|
meta.strides =
|
|
DDim(output_stride.data(), static_cast<int>(output_stride.size()));
|
|
out->set_meta(meta);
|
|
out->ResetHolder(input.Holder());
|
|
out->ShareInplaceVersionCounterWith(input);
|
|
}
|
|
|
|
} // namespace phi
|
|
|
|
PD_REGISTER_KERNEL_FOR_ALL_BACKEND_DTYPE(slice,
|
|
STRIDED,
|
|
phi::SliceStridedKernel) {}
|