461 lines
14 KiB
Plaintext
461 lines
14 KiB
Plaintext
// 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/backends/gpu/gpu_context.h"
|
|
#ifndef PADDLE_WITH_XPU_KP
|
|
#endif
|
|
#include "paddle/phi/core/kernel_registry.h"
|
|
#include "paddle/phi/kernels/impl/elementwise_kernel_impl.h"
|
|
#include "paddle/phi/kernels/legacy/elementwise_add_kernel.h"
|
|
#include "paddle/phi/kernels/legacy/elementwise_divide_kernel.h"
|
|
#include "paddle/phi/kernels/legacy/elementwise_kernel.h"
|
|
#include "paddle/phi/kernels/legacy/elementwise_multiply_kernel.h"
|
|
#include "paddle/phi/kernels/legacy/elementwise_subtract_kernel.h"
|
|
|
|
namespace phi {
|
|
|
|
template <typename T, typename Context>
|
|
PADDLE_API void SubtractKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
if (out->numel() == 0) {
|
|
dev_ctx.template Alloc<T>(out);
|
|
return;
|
|
}
|
|
phi::SubtractRawKernel<T, Context>(dev_ctx, x, y, -1, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void MultiplyKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
if (x.numel() == 0 || y.numel() == 0) {
|
|
dev_ctx.template Alloc<T>(out);
|
|
return;
|
|
}
|
|
phi::MultiplyRawKernel<T, Context>(dev_ctx, x, y, -1, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void DivideKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
if (x.numel() == 0 || y.numel() == 0) {
|
|
dev_ctx.template Alloc<T>(out);
|
|
return;
|
|
}
|
|
phi::DivideRawKernel<T, Context>(dev_ctx, x, y, -1, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void MultiPrecisionAddKernelImpl(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
std::vector<const DenseTensor*> inputs = {&x, &y};
|
|
std::vector<DenseTensor*> outputs = {out};
|
|
if (y.dtype() == phi::DataType::BFLOAT16) {
|
|
funcs::BroadcastKernel<T>(
|
|
dev_ctx,
|
|
inputs,
|
|
&outputs,
|
|
funcs::MultiPrecisionAddFunctor<T, phi::bfloat16>(),
|
|
-1);
|
|
} else if (y.dtype() == phi::DataType::FLOAT16) {
|
|
funcs::BroadcastKernel<T>(
|
|
dev_ctx,
|
|
inputs,
|
|
&outputs,
|
|
funcs::MultiPrecisionAddFunctor<T, phi::float16>(),
|
|
-1);
|
|
} else {
|
|
PADDLE_THROW(common::errors::InvalidArgument(
|
|
"Unsupported x dtype:%s, y dtype:%s for add(x, y) operation",
|
|
phi::DataTypeToString(x.type()),
|
|
phi::DataTypeToString(y.type())));
|
|
}
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void AddKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
#ifdef PADDLE_WITH_CUDA
|
|
if (x.dtype() == DataType::FLOAT32 &&
|
|
(y.dtype() == DataType::FLOAT16 || y.dtype() == DataType::BFLOAT16)) {
|
|
if (x.numel() == 0 || y.numel() == 0) {
|
|
dev_ctx.template Alloc<float>(out);
|
|
return;
|
|
}
|
|
MultiPrecisionAddKernelImpl<float, Context>(dev_ctx, x, y, out);
|
|
return;
|
|
}
|
|
#endif
|
|
if (x.numel() == 0 || y.numel() == 0) {
|
|
dev_ctx.template Alloc<T>(out);
|
|
return;
|
|
}
|
|
phi::AddRawKernel<T, Context>(dev_ctx, x, y, -1, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void GradAddKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
phi::AddRawKernel<T>(dev_ctx, x, y, -1, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void MaximumKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
int axis = -1;
|
|
MaximumRawKernel<T>(dev_ctx, x, y, axis, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void MinimumKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
int axis = -1;
|
|
MinimumRawKernel<T>(dev_ctx, x, y, axis, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void RemainderKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
int axis = -1;
|
|
RemainderRawKernel<T>(dev_ctx, x, y, axis, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void FloorDivideKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
int axis = -1;
|
|
FloorDivideRawKernel<T>(dev_ctx, x, y, axis, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void TruncDivideKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
int axis = -1;
|
|
std::vector<const DenseTensor*> inputs = {&x, &y};
|
|
std::vector<DenseTensor*> outputs = {out};
|
|
dev_ctx.template Alloc<T>(out);
|
|
funcs::BroadcastKernel<T>(
|
|
dev_ctx, inputs, &outputs, funcs::TruncDivideFunctor<T>(), axis);
|
|
}
|
|
|
|
// Create the definition of Heaviside
|
|
template <typename T, typename Context>
|
|
void HeavisideKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
std::vector<const DenseTensor*> inputs = {&x, &y};
|
|
std::vector<DenseTensor*> outputs = {out};
|
|
dev_ctx.template Alloc<T>(out);
|
|
funcs::BroadcastKernel<T>(
|
|
dev_ctx, inputs, &outputs, funcs::ElementwiseHeavisideFunctor<T>());
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void ElementwisePowKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
int axis = -1;
|
|
ElementwisePowRawKernel<T>(dev_ctx, x, y, axis, out);
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void CopySignKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
if (out->numel() == 0) {
|
|
dev_ctx.template Alloc<T>(out);
|
|
return;
|
|
}
|
|
std::vector<const DenseTensor*> inputs = {&x, &y};
|
|
std::vector<DenseTensor*> outputs = {out};
|
|
dev_ctx.template Alloc<T>(out);
|
|
funcs::BroadcastKernel<T>(
|
|
dev_ctx, inputs, &outputs, funcs::CopySignFunctor<T>());
|
|
}
|
|
|
|
template <typename T, typename Context>
|
|
void NextafterKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const DenseTensor& y,
|
|
DenseTensor* out) {
|
|
if (x.numel() == 0 || y.numel() == 0) {
|
|
dev_ctx.template Alloc<T>(out);
|
|
return;
|
|
}
|
|
std::vector<const DenseTensor*> inputs = {&x, &y};
|
|
std::vector<DenseTensor*> outputs = {out};
|
|
dev_ctx.template Alloc<T>(out);
|
|
funcs::BroadcastKernel<T>(
|
|
dev_ctx, inputs, &outputs, funcs::NextafterFunctor<T>());
|
|
}
|
|
#ifdef _WIN32
|
|
#define INSTANTIATE_ADD_KERNEL(type, context) \
|
|
template PADDLE_API void AddKernel<type, context>( \
|
|
const context&, const DenseTensor&, const DenseTensor&, DenseTensor*);
|
|
INSTANTIATE_ADD_KERNEL(float, GPUContext)
|
|
INSTANTIATE_ADD_KERNEL(double, GPUContext)
|
|
INSTANTIATE_ADD_KERNEL(phi::float16, GPUContext)
|
|
INSTANTIATE_ADD_KERNEL(phi::bfloat16, GPUContext)
|
|
INSTANTIATE_ADD_KERNEL(phi::complex64, GPUContext)
|
|
INSTANTIATE_ADD_KERNEL(phi::complex128, GPUContext)
|
|
#endif
|
|
} // namespace phi
|
|
|
|
#if defined(PADDLE_WITH_CUDA) || defined(PADDLE_WITH_HIP)
|
|
|
|
PD_REGISTER_KERNEL(maximum,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::MaximumKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
int64_t,
|
|
phi::float16,
|
|
phi::bfloat16) {}
|
|
PD_REGISTER_KERNEL(minimum,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::MinimumKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
int64_t,
|
|
phi::float16,
|
|
phi::bfloat16) {}
|
|
PD_REGISTER_KERNEL(remainder,
|
|
GPU,
|
|
ALL_LAYOUT,
|
|
phi::RemainderKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
int64_t,
|
|
phi::float16,
|
|
phi::complex64,
|
|
phi::complex128,
|
|
phi::bfloat16) {}
|
|
PD_REGISTER_KERNEL(floor_divide,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::FloorDivideKernel,
|
|
uint8_t,
|
|
int8_t,
|
|
int16_t,
|
|
int,
|
|
int64_t,
|
|
float,
|
|
double,
|
|
phi::float16,
|
|
phi::bfloat16) {}
|
|
PD_REGISTER_KERNEL(trunc_divide,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::TruncDivideKernel,
|
|
uint8_t,
|
|
int8_t,
|
|
int16_t,
|
|
int,
|
|
int64_t,
|
|
float,
|
|
double,
|
|
phi::dtype::float16,
|
|
phi::dtype::bfloat16) {}
|
|
PD_REGISTER_KERNEL(elementwise_pow,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::ElementwisePowKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
int64_t,
|
|
phi::float16,
|
|
phi::bfloat16,
|
|
phi::complex64,
|
|
phi::complex128) {}
|
|
PD_REGISTER_KERNEL(copysign,
|
|
GPU,
|
|
ALL_LAYOUT,
|
|
phi::CopySignKernel,
|
|
bool,
|
|
uint8_t,
|
|
int8_t,
|
|
int16_t,
|
|
int,
|
|
int64_t,
|
|
float,
|
|
double,
|
|
phi::float16,
|
|
phi::bfloat16) {}
|
|
PD_REGISTER_KERNEL(
|
|
nextafter, GPU, ALL_LAYOUT, phi::NextafterKernel, float, double) {}
|
|
|
|
#endif
|
|
|
|
#ifdef PADDLE_WITH_XPU_KP
|
|
PD_REGISTER_KERNEL(maximum, KPS, ALL_LAYOUT, phi::MaximumKernel, float) {}
|
|
PD_REGISTER_KERNEL(minimum, KPS, ALL_LAYOUT, phi::MinimumKernel, float) {}
|
|
PD_REGISTER_KERNEL(divide, KPS, ALL_LAYOUT, phi::DivideKernel, float) {}
|
|
PD_REGISTER_KERNEL(multiply, KPS, ALL_LAYOUT, phi::MultiplyKernel, float) {}
|
|
PD_REGISTER_KERNEL(add, KPS, ALL_LAYOUT, phi::AddKernel, float) {}
|
|
PD_REGISTER_KERNEL(subtract, KPS, ALL_LAYOUT, phi::SubtractKernel, float) {}
|
|
PD_REGISTER_KERNEL(floor_divide, KPS, ALL_LAYOUT, phi::FloorDivideKernel, int) {
|
|
}
|
|
PD_REGISTER_KERNEL(
|
|
elementwise_pow, KPS, ALL_LAYOUT, phi::ElementwisePowKernel, float) {}
|
|
|
|
#else
|
|
using float16 = phi::float16;
|
|
using bfloat16 = phi::bfloat16;
|
|
using complex64 = phi::complex64;
|
|
using complex128 = phi::complex128;
|
|
|
|
PD_REGISTER_KERNEL(fmax,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::FMaxKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
float16,
|
|
bfloat16,
|
|
int64_t) {}
|
|
|
|
PD_REGISTER_KERNEL(fmin,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::FMinKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
float16,
|
|
bfloat16,
|
|
int64_t) {}
|
|
|
|
PD_REGISTER_KERNEL(heaviside,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::HeavisideKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
float16,
|
|
bfloat16,
|
|
int64_t) {}
|
|
|
|
PD_REGISTER_KERNEL(add,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::AddKernel,
|
|
float,
|
|
double,
|
|
int16_t,
|
|
int,
|
|
bool,
|
|
uint8_t,
|
|
int8_t,
|
|
int64_t,
|
|
float16,
|
|
bfloat16,
|
|
complex64,
|
|
complex128) {}
|
|
|
|
PD_REGISTER_KERNEL(grad_add,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::GradAddKernel,
|
|
float,
|
|
double,
|
|
int16_t,
|
|
int,
|
|
bool,
|
|
uint8_t,
|
|
int8_t,
|
|
int64_t,
|
|
float16,
|
|
bfloat16,
|
|
complex64,
|
|
complex128) {}
|
|
|
|
PD_REGISTER_KERNEL(divide,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::DivideKernel,
|
|
float,
|
|
double,
|
|
int8_t,
|
|
uint8_t,
|
|
int16_t,
|
|
int,
|
|
int64_t,
|
|
bool,
|
|
float16,
|
|
bfloat16,
|
|
complex64,
|
|
complex128) {}
|
|
|
|
PD_REGISTER_KERNEL(multiply,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::MultiplyKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
int64_t,
|
|
bool,
|
|
float16,
|
|
complex64,
|
|
complex128,
|
|
bfloat16) {}
|
|
|
|
PD_REGISTER_KERNEL(subtract,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::SubtractKernel,
|
|
float,
|
|
double,
|
|
int16_t,
|
|
int,
|
|
int64_t,
|
|
float16,
|
|
bfloat16,
|
|
complex64,
|
|
complex128) {}
|
|
|
|
#endif
|