wehub-resource-sync
80 lines
2.4 KiB
Plaintext
80 lines
2.4 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/kernels/erfinv_kernel.h"
|
|
|
|
#include <limits>
|
|
|
|
#include "paddle/phi/backends/gpu/gpu_context.h"
|
|
#include "paddle/phi/core/kernel_registry.h"
|
|
#include "paddle/phi/kernels/funcs/elementwise_base.h"
|
|
|
|
namespace phi {
|
|
|
|
template <typename T>
|
|
struct ErfinvFunctor {
|
|
HOSTDEVICE inline T operator()(const T x) const {
|
|
// erfinv is only defined on [-1, 1]; align with PyTorch/scipy by
|
|
// returning NaN for |x| > 1 (CUDA erfinv returns +/-inf otherwise).
|
|
if (x > static_cast<T>(1) || x < static_cast<T>(-1)) {
|
|
return std::numeric_limits<T>::quiet_NaN();
|
|
}
|
|
return erfinv(x);
|
|
}
|
|
};
|
|
template <>
|
|
struct ErfinvFunctor<float16> {
|
|
HOSTDEVICE inline float16 operator()(const float16 x) const {
|
|
auto x_ = static_cast<float>(x);
|
|
if (x_ > 1.0f || x_ < -1.0f) {
|
|
return std::numeric_limits<float16>::quiet_NaN();
|
|
}
|
|
return static_cast<float16>(erfinv(x_));
|
|
}
|
|
};
|
|
|
|
template <>
|
|
struct ErfinvFunctor<bfloat16> {
|
|
HOSTDEVICE inline bfloat16 operator()(const bfloat16 x) const {
|
|
auto x_ = static_cast<float>(x);
|
|
if (x_ > 1.0f || x_ < -1.0f) {
|
|
return std::numeric_limits<bfloat16>::quiet_NaN();
|
|
}
|
|
return static_cast<bfloat16>(erfinv(x_));
|
|
}
|
|
};
|
|
template <typename T, typename Context>
|
|
void ErfinvKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
DenseTensor* out) {
|
|
dev_ctx.template Alloc<T>(out);
|
|
if (out && out->numel() == 0) {
|
|
return;
|
|
}
|
|
std::vector<const DenseTensor*> ins = {&x};
|
|
std::vector<DenseTensor*> outs = {out};
|
|
funcs::ElementwiseKernel<T>(dev_ctx, ins, &outs, ErfinvFunctor<T>());
|
|
}
|
|
|
|
} // namespace phi
|
|
|
|
PD_REGISTER_KERNEL(erfinv,
|
|
GPU,
|
|
ALL_LAYOUT,
|
|
phi::ErfinvKernel,
|
|
float,
|
|
double,
|
|
phi::float16,
|
|
phi::bfloat16) {}
|