• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /* Copyright 2015 The TensorFlow Authors. All Rights Reserved.
2 
3 Licensed under the Apache License, Version 2.0 (the "License");
4 you may not use this file except in compliance with the License.
5 You may obtain a copy of the License at
6 
7     http://www.apache.org/licenses/LICENSE-2.0
8 
9 Unless required by applicable law or agreed to in writing, software
10 distributed under the License is distributed on an "AS IS" BASIS,
11 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12 See the License for the specific language governing permissions and
13 limitations under the License.
14 ==============================================================================*/
15 
16 // See docs in ../ops/math_ops.cc.
17 
18 #define EIGEN_USE_THREADS
19 
20 #include "tensorflow/core/framework/op.h"
21 #include "tensorflow/core/framework/op_kernel.h"
22 #include "tensorflow/core/framework/type_traits.h"
23 #include "tensorflow/core/framework/types.h"
24 #include "tensorflow/core/kernels/meta_support.h"
25 #include "tensorflow/core/kernels/quantization_utils.h"
26 #include "tensorflow/core/lib/core/errors.h"
27 
28 namespace {
29 enum {
30   QUANTIZE_MODE_MIN_COMBINED,
31   QUANTIZE_MODE_MIN_FIRST,
32   QUANTIZE_MODE_SCALED,
33 };
34 }  // namespace
35 
36 namespace tensorflow {
37 
38 typedef Eigen::ThreadPoolDevice CPUDevice;
39 
40 template <typename Device, typename T>
41 class DequantizeOp : public OpKernel {
42  public:
DequantizeOp(OpKernelConstruction * ctx)43   explicit DequantizeOp(OpKernelConstruction* ctx) : OpKernel(ctx) {
44     half_range_ = !std::is_signed<T>::value
45                       ? 0.0f
46                       : (static_cast<float>(std::numeric_limits<T>::max()) -
47                          std::numeric_limits<T>::min() + 1) /
48                             2.0f;
49     string mode_string;
50     OP_REQUIRES_OK(ctx, ctx->GetAttr("mode", &mode_string));
51     OP_REQUIRES(ctx,
52                 (mode_string == "MIN_COMBINED" || mode_string == "MIN_FIRST" ||
53                  mode_string == "SCALED"),
54                 errors::InvalidArgument("Mode string must be 'MIN_COMBINED',"
55                                         " 'MIN_FIRST', or 'SCALED', is '" +
56                                         mode_string + "'"));
57     if (mode_string == "MIN_COMBINED") {
58       mode_ = QUANTIZE_MODE_MIN_COMBINED;
59     } else if (mode_string == "MIN_FIRST") {
60       mode_ = QUANTIZE_MODE_MIN_FIRST;
61     } else if (mode_string == "SCALED") {
62       mode_ = QUANTIZE_MODE_SCALED;
63     }
64   }
65 
Compute(OpKernelContext * ctx)66   void Compute(OpKernelContext* ctx) override {
67     const Tensor& input = ctx->input(0);
68     const float min_range = ctx->input(1).flat<float>()(0);
69     const float max_range = ctx->input(2).flat<float>()(0);
70 
71     Tensor* output = nullptr;
72     OP_REQUIRES_OK(ctx, ctx->allocate_output(0, input.shape(), &output));
73     if (mode_ == QUANTIZE_MODE_MIN_COMBINED) {
74       const float scale_factor =
75           (max_range - min_range) /
76           (static_cast<float>(std::numeric_limits<T>::max()) -
77            std::numeric_limits<T>::min());
78 
79       float* out_ptr = output->flat<float>().data();
80       const T* in_ptr = input.flat<T>().data();
81 
82       const int64 num_elements = input.NumElements();
83       for (int i = 0; i < num_elements; ++i) {
84         out_ptr[i] =
85             ((static_cast<int>(in_ptr[i]) + half_range_) * scale_factor) +
86             min_range;
87       }
88     } else if (mode_ == QUANTIZE_MODE_MIN_FIRST) {
89       if (meta::IsSupportedAndEnabled() && std::is_same<T, quint8>()) {
90         auto input_ui8_array = input.flat<quint8>();
91         meta::Dequantize(ctx, input_ui8_array.data(), input_ui8_array.size(),
92                          min_range, max_range, output->flat<float>().data());
93       } else {
94         QuantizedTensorToFloatInPlaceUsingEigen<T>(
95             ctx->template eigen_device<Device>(), input, min_range, max_range,
96             output);
97       }
98     } else if (mode_ == QUANTIZE_MODE_SCALED) {
99       const float scale_factor =
100           std::numeric_limits<T>::min() == 0
101               ? (max_range / std::numeric_limits<T>::max())
102               : std::max(min_range / std::numeric_limits<T>::min(),
103                          max_range / std::numeric_limits<T>::max());
104       float* out_ptr = output->flat<float>().data();
105       const T* in_ptr = input.flat<T>().data();
106       const int64 num_elements = input.NumElements();
107       for (int64 i = 0; i < num_elements; ++i) {
108         out_ptr[i] = static_cast<int>(in_ptr[i]) * scale_factor;
109       }
110     }
111   }
112 
113  private:
114   float half_range_;
115   int mode_;
116 };
117 
118 REGISTER_KERNEL_BUILDER(
119     Name("Dequantize").Device(DEVICE_CPU).TypeConstraint<quint8>("T"),
120     DequantizeOp<CPUDevice, quint8>);
121 REGISTER_KERNEL_BUILDER(
122     Name("Dequantize").Device(DEVICE_CPU).TypeConstraint<qint8>("T"),
123     DequantizeOp<CPUDevice, qint8>);
124 REGISTER_KERNEL_BUILDER(
125     Name("Dequantize").Device(DEVICE_CPU).TypeConstraint<quint16>("T"),
126     DequantizeOp<CPUDevice, quint16>);
127 REGISTER_KERNEL_BUILDER(
128     Name("Dequantize").Device(DEVICE_CPU).TypeConstraint<qint16>("T"),
129     DequantizeOp<CPUDevice, qint16>);
130 
131 REGISTER_KERNEL_BUILDER(
132     Name("Dequantize").Device(DEVICE_CPU).TypeConstraint<qint32>("T"),
133     DequantizeOp<CPUDevice, qint32>);
134 
135 }  // namespace tensorflow
136