tensor.h

/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

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. */

#pragma once

#include <cstdint>
#include <cstring>
#include <memory>
#include <type_traits>
#include "paddle/framework/ddim.h"
#include "paddle/framework/enforce.h"
#include "paddle/memory/memory.h"
#include "paddle/platform/place.h"

namespace paddle {
namespace framework {

class Tensor {
 public:
  Tensor() : offset_(0) { numel_ = product(dims_); }

  Tensor& operator=(const Tensor& src) = delete;

  template <typename T>
  const T* data() const {
    CheckDimsValidity();
    return reinterpret_cast<const T*>(
        reinterpret_cast<uintptr_t>(holder_->Ptr()) + offset_);
  }

  template <typename T>
  T* mutable_data(DDim dims, paddle::platform::Place place) {
    set_dims(dims);
    return mutable_data<T>(place);
  }

  template <typename T>
  T* mutable_data(paddle::platform::Place place) {
    PADDLE_ENFORCE(numel_ > 0,
                   "Tensor::numel_ must be larger than zero to call "
                   "Tensor::mutable_data.");
    if (holder_ == nullptr ||
        !(holder_->Place() ==
          place) /* some versions of boost::variant don't have operator!= */
        || holder_->Size() < numel_ * sizeof(T) + offset_) {
      holder_.reset(new PlaceholderImpl<T>(place, numel_ * sizeof(T)));
      offset_ = 0;
    }
    return reinterpret_cast<T*>(reinterpret_cast<uintptr_t>(holder_->Ptr()) +
                                offset_);
  }

  void ShareDataFrom(const Tensor& src) {
    src.CheckDimsValidity();
    holder_ = src.holder_;
    dims_ = src.dims();
    numel_ = src.numel_;
    offset_ = src.offset_;
  }

  void CopyFrom(const Tensor& src, paddle::platform::Place dst_place) {
    src.CheckDimsValidity();
    size_t size = src.numel_ * src.holder_->TypeSize();
    holder_.reset(src.holder_->Clone(src.offset_, size, dst_place));
    dims_ = src.dims();
    numel_ = src.numel_;
    offset_ = 0;
  }

  Tensor Slice(const int& begin_idx, const int& end_idx) const {
    CheckDimsValidity();
    PADDLE_ENFORCE(begin_idx >= 0 && end_idx <= dims_[0],
                   "Slice index is less than zero or out of bound.");
    PADDLE_ENFORCE(begin_idx < end_idx,
                   "Begin index must be less than end index.");
    PADDLE_ENFORCE(dims_[0] != 1, "Can not slice a tensor with dims_[0] = 1.");
    std::vector<int> d = vectorize(dims_);
    int base = 1;
    for (size_t i = 1; i < d.size(); ++i) {
      base *= d[i];
    }
    Tensor dst;
    dst.holder_ = holder_;
    DDim dst_dims = dims_;
    dst_dims[0] = end_idx - begin_idx;
    dst.set_dims(dst_dims);
    dst.offset_ = offset_ + begin_idx * base * holder_->TypeSize();
    return dst;
  }

  void set_dims(const DDim& dims) {
    if (dims == dims_) {
      return;
    }
    dims_ = dims;
    numel_ = product(dims_);
    return;
  }

  DDim dims() const { return dims_; }

 private:
  // Placeholder hides type T, so it doesn't appear as a template
  // parameter of Variable.
  struct Placeholder {
    virtual ~Placeholder() {}
    virtual void* Ptr() const = 0;
    virtual paddle::platform::Place Place() const = 0;
    virtual size_t Size() const = 0;
    virtual size_t TypeSize() const = 0;
    virtual Placeholder* Clone(size_t begin, size_t size,
                               paddle::platform::Place place) const = 0;
  };

  template <typename T>
  struct PlaceholderImpl : public Placeholder {
   private:
    class Deleter {
     public:
      Deleter(platform::Place place) : place_(place) {}
      void operator()(T* ptr) {
        paddle::memory::Free(place_, static_cast<void*>(ptr));
      }

     private:
      paddle::platform::Place place_;
    };

   public:
    PlaceholderImpl(paddle::platform::Place place, size_t size)
        : ptr_(static_cast<T*>(paddle::memory::Alloc(place, size)),
               Deleter(place)),
          place_(place),
          size_(size) {}

    virtual void* Ptr() const { return static_cast<void*>(ptr_.get()); }
    virtual size_t Size() const { return size_; }
    virtual paddle::platform::Place Place() const { return place_; }
    virtual size_t TypeSize() const { return sizeof(T); }
    // TODO: Clone only support CPU now. GPU support is needed.
    virtual Placeholder* Clone(size_t begin, size_t size,
                               paddle::platform::Place place) const {
      PADDLE_ENFORCE(paddle::platform::is_cpu_place(place_) &&
                         paddle::platform::is_cpu_place(place),
                     "PlaceholderImpl::Clone only support CPU now.");
      PlaceholderImpl<T>* dst = new PlaceholderImpl<T>(place, size);
      void* begin_ptr =
          reinterpret_cast<void*>(reinterpret_cast<uintptr_t>(Ptr()) + begin);
      memcpy(dst->Ptr(), begin_ptr, size);
      return dst;
    }

    std::unique_ptr<T, Deleter> ptr_;
    paddle::platform::Place place_;  // record the place of ptr_.
    size_t size_;                    // size of the memory block.
  };

  inline void CheckDimsValidity() {
    PADDLE_ENFORCE(holder_ != nullptr,
                   "Tenosr holds no memory. Call Tensor::mutable_data first.");
    PADDLE_ENFORCE(holder_->Size() > numel_ * sizeof(T) + offset_,
                   "Tensor's dims_ is out of bound. Call Tensor::mutable_data "
                   "first to re-allocate memory.");
  }

  std::shared_ptr<Placeholder> holder_;  // holds the memory block if allocated.
  DDim dims_;
  int numel_;      // cache of `product(dims_)`
  size_t offset_;  // marks the begin of tensor data area.
};

}  // namespace framework
}  // namespace paddle