wehub-resource-sync
110 lines
3.5 KiB
C++
110 lines
3.5 KiB
C++
// Copyright (c) 2022 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/gather_kernel.h"
|
|
|
|
#include "paddle/phi/backends/xpu/enforce_xpu.h"
|
|
#include "paddle/phi/core/kernel_registry.h"
|
|
#include "paddle/phi/kernels/full_kernel.h"
|
|
namespace phi {
|
|
|
|
template <typename T, typename Context>
|
|
void GatherGradKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& index,
|
|
const DenseTensor& out_grad,
|
|
const Scalar& axis,
|
|
DenseTensor* x_grad) {
|
|
auto axis_v = axis.to<int64_t>();
|
|
if (axis_v < 0) {
|
|
axis_v += static_cast<int64_t>(x.dims().size());
|
|
}
|
|
|
|
const auto& index_type = index.dtype();
|
|
|
|
if (out_grad.numel() == 0) {
|
|
if (x_grad) {
|
|
Full<T, Context>(dev_ctx, x_grad->dims(), 0, x_grad);
|
|
}
|
|
return;
|
|
}
|
|
|
|
const auto index_dims = index.dims();
|
|
if (index_dims.size() == 2) {
|
|
PADDLE_ENFORCE_EQ(
|
|
index_dims[1],
|
|
1,
|
|
common::errors::InvalidArgument(
|
|
"The last dim of index should be 1 when it is 2D, but we get %d",
|
|
index_dims[1]));
|
|
} else {
|
|
PADDLE_ENFORCE_EQ(
|
|
index_dims.size() == 1 || index_dims.size() == 0,
|
|
true,
|
|
common::errors::InvalidArgument(
|
|
"The index should be 0D or 1D, when it is not 2D, but we get %d",
|
|
index_dims.size()));
|
|
}
|
|
std::vector<int64_t> xshape(x_grad->dims().size());
|
|
for (int i = 0; i < x_grad->dims().size(); ++i) {
|
|
xshape[i] = x_grad->dims()[i];
|
|
}
|
|
|
|
dev_ctx.template Alloc<T>(x_grad);
|
|
using XPUType = typename XPUTypeTrait<T>::Type;
|
|
|
|
int r = 0;
|
|
if (index_type == DataType::INT32) {
|
|
r = xpu::gather_grad<XPUType, int>(
|
|
dev_ctx.x_context(),
|
|
reinterpret_cast<const XPUType*>(out_grad.data<T>()),
|
|
index.data<int>(),
|
|
reinterpret_cast<XPUType*>(x_grad->data<T>()),
|
|
xshape,
|
|
index.dims().size() == 0 ? 1 : index.dims()[0],
|
|
axis_v,
|
|
false);
|
|
} else if (index_type == DataType::INT64) {
|
|
r = xpu::gather_grad<XPUType, int64_t>(
|
|
dev_ctx.x_context(),
|
|
reinterpret_cast<const XPUType*>(out_grad.data<T>()),
|
|
index.data<int64_t>(),
|
|
reinterpret_cast<XPUType*>(x_grad->data<T>()),
|
|
xshape,
|
|
index.dims().size() == 0 ? 1 : index.dims()[0],
|
|
axis_v,
|
|
false);
|
|
} else {
|
|
PADDLE_THROW(common::errors::InvalidArgument(
|
|
"Unsupported index type, expected int32 or int64, but got type %s",
|
|
DataTypeToString(index_type)));
|
|
}
|
|
PADDLE_ENFORCE_XDNN_SUCCESS(r, "gather_grad");
|
|
}
|
|
|
|
} // namespace phi
|
|
|
|
PD_REGISTER_KERNEL(gather_grad,
|
|
XPU,
|
|
ALL_LAYOUT,
|
|
phi::GatherGradKernel,
|
|
float,
|
|
phi::float16,
|
|
phi::bfloat16,
|
|
int8_t,
|
|
int16_t,
|
|
int32_t,
|
|
int64_t,
|
|
bool) {}
|