chromium/third_party/tflite/src/tensorflow/lite/kernels/internal/reference/div.h

/* Copyright 2020 The TensorFlow 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.
==============================================================================*/
#ifndef TENSORFLOW_LITE_KERNELS_INTERNAL_REFERENCE_DIV_H_
#define TENSORFLOW_LITE_KERNELS_INTERNAL_REFERENCE_DIV_H_

#include <algorithm>

#include "tensorflow/lite/kernels/internal/common.h"

namespace tflite {

namespace reference_ops {

template <typename T>
inline void DivCheckArithmeticParams(const ArithmeticParams& params) { … }

// Element-wise div that can often be used for inner loop of broadcast Div as
// well as the non-broadcast Div.
template <typename T>
inline void DivElementwise(int size, const ArithmeticParams& params,
                           const T* input1_data, const T* input2_data,
                           T* output_data) { … }

inline void Div(const ArithmeticParams& params,
                const RuntimeShape& input1_shape, const uint8_t* input1_data,
                const RuntimeShape& input2_shape, const uint8_t* input2_data,
                const RuntimeShape& output_shape, uint8_t* output_data) { … }

inline void Div(const ArithmeticParams& params,
                const RuntimeShape& input1_shape, const int8_t* input1_data,
                const RuntimeShape& input2_shape, const int8_t* input2_data,
                const RuntimeShape& output_shape, int8_t* output_data) { … }

template <typename T, int N = 5>
inline void BroadcastDivSlowQuantized(
    const ArithmeticParams& params, const RuntimeShape& unextended_input1_shape,
    const T* input1_data, const RuntimeShape& unextended_input2_shape,
    const T* input2_data, const RuntimeShape& unextended_output_shape,
    T* output_data) { … }

template <int N = 5>
inline void BroadcastDivSlow(const ArithmeticParams& params,
                             const RuntimeShape& unextended_input1_shape,
                             const uint8_t* input1_data,
                             const RuntimeShape& unextended_input2_shape,
                             const uint8_t* input2_data,
                             const RuntimeShape& unextended_output_shape,
                             uint8_t* output_data) { … }

template <int N = 5>
inline void BroadcastDivSlow(const ArithmeticParams& params,
                             const RuntimeShape& unextended_input1_shape,
                             const int8_t* input1_data,
                             const RuntimeShape& unextended_input2_shape,
                             const int8_t* input2_data,
                             const RuntimeShape& unextended_output_shape,
                             int8_t* output_data) { … }

// TODO(jiawen): We can implement BroadcastDiv on buffers of arbitrary
// dimensionality if the runtime code does a single loop over one dimension
// that handles broadcasting as the base case. The code generator would then
// generate max(D1, D2) nested for loops.
template <typename T, int N = 5>
void BroadcastDivSlow(const ArithmeticParams& params,
                      const RuntimeShape& unextended_input1_shape,
                      const T* input1_data,
                      const RuntimeShape& unextended_input2_shape,
                      const T* input2_data,
                      const RuntimeShape& unextended_output_shape,
                      T* output_data) { … }

template <typename T>
inline void Div(const ArithmeticParams& params,
                const RuntimeShape& input1_shape, const T* input1_data,
                const RuntimeShape& input2_shape, const T* input2_data,
                const RuntimeShape& output_shape, T* output_data) { … }

}  // namespace reference_ops
}  // namespace tflite

#endif  // TENSORFLOW_LITE_KERNELS_INTERNAL_REFERENCE_DIV_H_