135 lines
4.5 KiB
C++
135 lines
4.5 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/transpose_grad_kernel.h"
|
|
|
|
#include "paddle/phi/backends/xpu/enforce_xpu.h"
|
|
#include "paddle/phi/core/kernel_registry.h"
|
|
#include "paddle/phi/kernels/complex_kernel.h"
|
|
|
|
namespace phi {
|
|
|
|
template <typename T, typename Context>
|
|
void TransposeGradKernel(const Context& dev_ctx,
|
|
const DenseTensor& out_grad,
|
|
const std::vector<int>& axis,
|
|
DenseTensor* x_grad) {
|
|
using XPUType = typename XPUTypeTrait<T>::Type;
|
|
dev_ctx.template Alloc<T>(x_grad);
|
|
if (x_grad->numel() == 0) {
|
|
return;
|
|
}
|
|
|
|
size_t axis_size = axis.size();
|
|
if (axis_size == 0) {
|
|
Copy<Context>(dev_ctx, out_grad, dev_ctx.GetPlace(), false, x_grad);
|
|
return;
|
|
}
|
|
|
|
std::vector<int64_t> formatted_axis(axis.begin(), axis.end());
|
|
for (size_t i = 0; i < axis_size; i++) {
|
|
if (axis[i] < 0) {
|
|
formatted_axis[i] = axis[i] + axis_size;
|
|
}
|
|
}
|
|
|
|
std::vector<int64_t> reversed_axis(axis.begin(), axis.end());
|
|
for (size_t i = 0; i < axis_size; i++) {
|
|
reversed_axis[formatted_axis[i]] = i;
|
|
}
|
|
|
|
std::vector<int64_t> out_grad_dim_vec = vectorize<int64_t>(out_grad.dims());
|
|
int r = xpu::transpose<XPUType>(
|
|
dev_ctx.x_context(),
|
|
reinterpret_cast<const XPUType*>(out_grad.data<T>()),
|
|
reinterpret_cast<XPUType*>(x_grad->data<T>()),
|
|
out_grad_dim_vec,
|
|
reversed_axis);
|
|
PADDLE_ENFORCE_XDNN_SUCCESS(r, "transpose_grad");
|
|
}
|
|
|
|
#ifdef PADDLE_WITH_XPU_FFT
|
|
template <>
|
|
void TransposeGradKernel<phi::complex64, XPUContext>(
|
|
const XPUContext& dev_ctx,
|
|
const DenseTensor& out_grad,
|
|
const std::vector<int>& axis,
|
|
DenseTensor* x_grad) {
|
|
using T = phi::complex64;
|
|
dev_ctx.template Alloc<T>(x_grad);
|
|
if (x_grad->numel() == 0) {
|
|
return;
|
|
}
|
|
|
|
size_t axis_size = axis.size();
|
|
if (axis_size == 0) {
|
|
Copy<XPUContext>(dev_ctx, out_grad, dev_ctx.GetPlace(), false, x_grad);
|
|
return;
|
|
}
|
|
|
|
std::vector<int64_t> formatted_axis(axis.begin(), axis.end());
|
|
for (size_t i = 0; i < axis_size; i++) {
|
|
if (axis[i] < 0) {
|
|
formatted_axis[i] = axis[i] + axis_size;
|
|
}
|
|
}
|
|
|
|
std::vector<int64_t> reversed_axis(axis.begin(), axis.end());
|
|
for (size_t i = 0; i < axis_size; i++) {
|
|
reversed_axis[formatted_axis[i]] = i;
|
|
}
|
|
|
|
// The current complex number implementation uses separate real/imaginary
|
|
// parts,resulting in redundant operations and performance
|
|
// penalties.Optimization should address this in future iterations.
|
|
DenseTensor real_out, imag_out;
|
|
real_out.Resize(x_grad->dims());
|
|
imag_out.Resize(x_grad->dims());
|
|
dev_ctx.template Alloc<float>(&real_out);
|
|
dev_ctx.template Alloc<float>(&imag_out);
|
|
const DenseTensor real = Real<T, XPUContext>(dev_ctx, out_grad);
|
|
const DenseTensor imag = Imag<T, XPUContext>(dev_ctx, out_grad);
|
|
std::vector<int64_t> out_grad_dim_vec = vectorize<int64_t>(out_grad.dims());
|
|
int r = xpu::transpose<float>(dev_ctx.x_context(),
|
|
real.data<float>(),
|
|
real_out.data<float>(),
|
|
out_grad_dim_vec,
|
|
reversed_axis);
|
|
PADDLE_ENFORCE_XDNN_SUCCESS(r, "transpose_grad");
|
|
r = xpu::transpose<float>(dev_ctx.x_context(),
|
|
imag.data<float>(),
|
|
imag_out.data<float>(),
|
|
out_grad_dim_vec,
|
|
reversed_axis);
|
|
PADDLE_ENFORCE_XDNN_SUCCESS(r, "transpose_grad");
|
|
phi::ComplexKernel<float>(dev_ctx, real_out, imag_out, x_grad);
|
|
}
|
|
#endif
|
|
} // namespace phi
|
|
|
|
PD_REGISTER_KERNEL(transpose_grad,
|
|
XPU,
|
|
ALL_LAYOUT,
|
|
phi::TransposeGradKernel,
|
|
float,
|
|
phi::float16,
|
|
phi::bfloat16,
|
|
#ifdef PADDLE_WITH_XPU_FFT
|
|
phi::complex64,
|
|
#endif
|
|
int64_t,
|
|
int,
|
|
bool) {
|
|
}
|