tensorflow--tensorflow/tensorflow/lite/simple_memory_arena.h

/* Copyright 2017 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_SIMPLE_MEMORY_ARENA_H_
#define TENSORFLOW_LITE_SIMPLE_MEMORY_ARENA_H_

#include <cstddef>
#include <cstdint>
#include <string>
#include <vector>

#include "tensorflow/lite/core/c/common.h"

namespace tflite {

// This little structure holds the offset and the size for a dynamic memory
// allocation in the memory arena as well as first_node and last_node that use
// corresponding tensor. It means that continuous part of memory with this size
// needs to be allocated before execution of operation in the first node and can
// be deallocated after execution of the operation in the last_node. When the
// arena is committed and the underlying buffer is set, the alloc can be
// resolved into an actual memory pointer.
struct ArenaAllocWithUsageInterval {
  ArenaAllocWithUsageInterval() { reset(); }

  size_t offset;
  size_t size;
  int32_t tensor;
  int32_t first_node;
  int32_t last_node;

  inline void reset() {
    offset = 0;
    size = 0;
    tensor = -1;
    first_node = -1;
    last_node = -1;
  }

  inline bool operator<(const ArenaAllocWithUsageInterval& other) const {
    return offset < other.offset;
  }
};

struct PointerAlignedPointerPair {
  char* pointer;
  char* aligned_pointer;
};

class ResizableAlignedBuffer {
 public:
  ResizableAlignedBuffer(size_t alignment, int subgraph_index)
      : buffer_{nullptr, nullptr},
        data_size_(0),
        alignment_(alignment),
        subgraph_index_(subgraph_index) {
    // To silence unused private member warning, only used with
    // TF_LITE_TENSORFLOW_PROFILER
    (void)subgraph_index_;
  }

  ~ResizableAlignedBuffer() { Release(); }

  // Resizes the buffer to make sure new_size bytes fit in the buffer. Keeps
  // alignment and any existing the data. Returns kTfLiteError without
  // modifying the existing allocation if the resize fails. Sets
  // `buffer_reallocated` when any external pointers into the data array need to
  // be adjusted (the buffer was moved).
  TfLiteStatus Resize(size_t new_size, bool* buffer_reallocated);
  // Releases any allocated memory.
  void Release();

  // Pointer to the data array.
  char* GetPtr() const { return buffer_.aligned_pointer; }
  // Size of the data array. Note: the allocated memory block might be larger
  // due to excess alignment requirements.
  size_t GetSize() const { return data_size_; }
  // Alignment of the data array.
  size_t GetAlignment() const { return alignment_; }

 private:
  ResizableAlignedBuffer(const ResizableAlignedBuffer&) = delete;
  ResizableAlignedBuffer& operator=(const ResizableAlignedBuffer&) = delete;
  ResizableAlignedBuffer(ResizableAlignedBuffer&&) = delete;
  ResizableAlignedBuffer& operator=(ResizableAlignedBuffer&&) = delete;

  PointerAlignedPointerPair buffer_;
  size_t data_size_;
  size_t alignment_;

  int subgraph_index_;
};

// This small class is responsible for allocating, deallocating and reusing
// dynamic memory from a common underlying buffer. The arena can be used in
// scenarios when the pattern of memory allocations and deallocations is
// repetitive, e.g. running NN inference in multiple iterations. Note that
// zero-sized allocations are explicitly allowed, and will resolve to null.
class SimpleMemoryArena {
 public:
  explicit SimpleMemoryArena(size_t arena_alignment, int subgraph_index = 0)
      : committed_(false),
        high_water_mark_(0),
        underlying_buffer_(arena_alignment, subgraph_index),
        active_allocs_() {}

  // Delete all allocs. This should be called when allocating the first node of
  // a subgraph.
  void ResetAllocs();

  // Delete all allocs which are deallocated before `node`. This should be
  // called before allocating tensors associated with a series of nodes. It
  // deletes allocs which are no longer required for allocating the next batch
  // of tensors. Not calling it will have no impact on the result but it may be
  // much slower.
  void PurgeActiveAllocs(int32_t node);

  // Delete all allocs which are allocated after `node`. This should be
  // called when resetting allocs after `node`. It  deletes allocs which are no
  // longer required for allocating the next batch of tensors. Not calling it
  // will have no impact on the result but it may be much slower.
  void PurgeAfter(int32_t node);

  // Calculate the active allocs at `node`. Call this if the active allocs at
  // `node` are unknown.
  void CalculateActiveAllocs(
      const std::vector<ArenaAllocWithUsageInterval>& allocs, int32_t node);

  // Schedule memory allocation for a tensor with a given size, assuming that it
  // needs to be allocated before the execution of first_node, and deallocated
  // after the execution of last_node.
  TfLiteStatus Allocate(TfLiteContext* context, size_t alignment, size_t size,
                        int32_t tensor, int32_t first_node, int32_t last_node,
                        ArenaAllocWithUsageInterval* new_alloc);

  TfLiteStatus Commit(bool* arena_reallocated);

  TfLiteStatus ResolveAlloc(TfLiteContext* context,
                            const ArenaAllocWithUsageInterval& alloc,
                            char** output_ptr);

  // This clears allocation details but does not release the underlying buffer.
  // New allocations should be committed & resolved before using this arena
  // again.
  TfLiteStatus ClearPlan();

  // This releases the underlying buffer but does not clear the allocation plan.
  // Since all associated pointers are invalidated, the arena cannot be used
  // again until Commit() is called & tensor allocations are resolved.
  TfLiteStatus ReleaseBuffer();

  size_t GetBufferSize() const { return underlying_buffer_.GetSize(); }

  std::intptr_t BasePointer() const {
    return reinterpret_cast<std::intptr_t>(underlying_buffer_.GetPtr());
  }

  // Dumps the memory allocation information of this memory arena (which could
  // be differentiated from others by the `name`) against the specified op node
  // execution plan (i.e. `execution_plan`) for the purpose of debugging.
  // Note: in order to have minimal binary increase caused by this debug info
  // dump implementation for the TfLite library, and allow users to plug-in
  // their own memory planner debugger, we have utilized weak symbols to meet
  // these two requirementsements. By default, there is no debugging info
  // dumped. To override this, provide a strong defintion of
  // tflite::DumpArenaInfo(...) whose weak defintion is in
  // simple_memory_arena.cc. TfLite provides a sample one as
  // "lite:simple_memory_arena_debug_dump". When this dep is added to the
  // program, calling this function will output information of this memory arena
  // about tenosrs and ops, such as memory arena utilization rate, live tensors
  // at each op etc.
  void DumpDebugInfo(const std::string& name,
                     const std::vector<int>& execution_plan) const;

 private:
  bool committed_;
  size_t high_water_mark_;
  ResizableAlignedBuffer underlying_buffer_;
  std::vector<ArenaAllocWithUsageInterval> active_allocs_;
};

}  // namespace tflite

#endif  // TENSORFLOW_LITE_SIMPLE_MEMORY_ARENA_H_