| /* |
| * Copyright (C) 2017 The Android Open Source Project |
| * |
| * 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. |
| */ |
| |
| // Contains the implementation of the operations. |
| |
| #define LOG_TAG "Operations" |
| |
| #include "Operations.h" |
| #include "CpuOperationUtils.h" |
| |
| #include "tensorflow/contrib/lite/kernels/internal/optimized/legacy_optimized_ops.h" |
| #include "tensorflow/contrib/lite/kernels/internal/reference/legacy_reference_ops.h" |
| |
| #include "Tracing.h" |
| |
| namespace android { |
| namespace nn { |
| |
| inline void get4DShape(const Shape& shapeIn, uint32_t* shapeOut) { |
| const int32_t kNumDims = 4; |
| int32_t numDims = static_cast<int32_t>(getNumberOfDimensions(shapeIn)); |
| for (int32_t i = 0; i < numDims; i++) { |
| shapeOut[i + kNumDims - numDims] = getSizeOfDimension(shapeIn, i); |
| } |
| for (int32_t i = 0; i < kNumDims - numDims; i++) { |
| shapeOut[i] = 1; |
| } |
| } |
| |
| inline void getBroadcastStrides(const Shape& shapeIn, uint32_t* stridesOut) { |
| const int32_t kNumDims = 4; |
| uint32_t dims[kNumDims]; |
| get4DShape(shapeIn, dims); |
| stridesOut[kNumDims - 1] = 1; |
| for (int32_t i = kNumDims - 2; i >= 0; i--) { |
| stridesOut[i] = stridesOut[i + 1] * dims[i + 1]; |
| } |
| for (int32_t i = kNumDims - 1; i >= 0; i--) { |
| if (dims[i] == 1) { |
| stridesOut[i] = 0; |
| } |
| } |
| } |
| |
| template <typename T> |
| inline bool broadcastOpBase(const T* in1, const Shape& shape1, const T* in2, const Shape& shape2, |
| T* out, const Shape& shapeOut, |
| std::function<T(const T&, const T&)> mathKernel) { |
| // extend to 4D |
| uint32_t outputDims[4], in1Strides[4], in2Strides[4]; |
| get4DShape(shapeOut, outputDims); |
| getBroadcastStrides(shape1, in1Strides); |
| getBroadcastStrides(shape2, in2Strides); |
| |
| T* outPtr = out; |
| for (uint32_t b = 0; b < outputDims[0]; b++) { |
| for (uint32_t h = 0; h < outputDims[1]; h++) { |
| for (uint32_t w = 0; w < outputDims[2]; w++) { |
| for (uint32_t c = 0; c < outputDims[3]; c++) { |
| T in1Val = in1[b * in1Strides[0] + h * in1Strides[1] + w * in1Strides[2] + |
| c * in1Strides[3]]; |
| T in2Val = in2[b * in2Strides[0] + h * in2Strides[1] + w * in2Strides[2] + |
| c * in2Strides[3]]; |
| *outPtr++ = mathKernel(in1Val, in2Val); |
| } |
| } |
| } |
| } |
| return true; |
| } |
| |
| #define ANDROID_NN_MACRO_DISPATCH(macro) \ |
| switch (activation) { \ |
| case (int32_t) FusedActivationFunc::NONE: \ |
| macro(kNone); \ |
| break; \ |
| case (int32_t) FusedActivationFunc::RELU: \ |
| macro(kRelu); \ |
| break; \ |
| case (int32_t) FusedActivationFunc::RELU1: \ |
| macro(kRelu1); \ |
| break; \ |
| case (int32_t) FusedActivationFunc::RELU6: \ |
| macro(kRelu6); \ |
| break; \ |
| default: \ |
| LOG(ERROR) << "Unsupported fused activation function type"; \ |
| return false; \ |
| } |
| |
| bool addFloat32(const float* in1, const Shape& shape1, |
| const float* in2, const Shape& shape2, |
| int32_t activation, |
| float* out, const Shape& shapeOut) { |
| NNTRACE_TRANS("addFloat32"); |
| bool needBroadcast = !SameShape(shape1, shape2); |
| |
| if (needBroadcast) { |
| NNTRACE_COMP_SWITCH("optimized_ops::BroadcastAdd"); |
| #define ANDROID_NN_BROADCAST_ADD(activation) \ |
| tflite::optimized_ops::BroadcastAdd<tflite::FusedActivationFunctionType::activation>( \ |
| in1, convertShapeToDims(shape1), \ |
| in2, convertShapeToDims(shape2), \ |
| out, convertShapeToDims(shapeOut)) |
| |
| ANDROID_NN_MACRO_DISPATCH(ANDROID_NN_BROADCAST_ADD) |
| #undef ANDROID_NN_BROADCAST_ADD |
| } else { |
| NNTRACE_COMP_SWITCH("optimized_ops::Add"); |
| #define ANDROID_NN_ADD(activation) \ |
| tflite::optimized_ops::Add<tflite::FusedActivationFunctionType::activation>( \ |
| in1, convertShapeToDims(shape1), \ |
| in2, convertShapeToDims(shape2), \ |
| out, convertShapeToDims(shapeOut)) |
| |
| ANDROID_NN_MACRO_DISPATCH(ANDROID_NN_ADD) |
| #undef ANDROID_NN_ADD |
| } |
| |
| return true; |
| } |
| |
| bool addQuant8(const uint8_t* in1, const Shape& shape1, |
| const uint8_t* in2, const Shape& shape2, |
| int32_t activation, |
| uint8_t* out, const Shape& shapeOut) { |
| NNTRACE_TRANS("addQuant8"); |
| bool needBroadcast = !SameShape(shape1, shape2); |
| |
| const int32_t input1_offset = -shape1.offset; |
| const int32_t input2_offset = -shape2.offset; |
| const int32_t output_offset = shapeOut.offset; |
| const int left_shift = 20; |
| const double twice_max_input_scale = 2 * std::max(shape1.scale, shape2.scale); |
| const double real_input1_multiplier = shape1.scale / twice_max_input_scale; |
| const double real_input2_multiplier = shape2.scale / twice_max_input_scale; |
| const double real_output_multiplier = |
| twice_max_input_scale / |
| ((1 << left_shift) * shapeOut.scale); |
| |
| int32_t input1_multiplier; |
| int32_t input1_shift; |
| if (!QuantizeMultiplierSmallerThanOne(real_input1_multiplier, |
| &input1_multiplier, &input1_shift)) { |
| return false; |
| } |
| int32_t input2_multiplier; |
| int32_t input2_shift; |
| if (!QuantizeMultiplierSmallerThanOne(real_input2_multiplier, |
| &input2_multiplier, &input2_shift)) { |
| return false; |
| } |
| int32_t output_multiplier; |
| int32_t output_shift; |
| if (!QuantizeMultiplierSmallerThanOne(real_output_multiplier, |
| &output_multiplier, &output_shift)) { |
| return false; |
| } |
| int32_t output_activation_min; |
| int32_t output_activation_max; |
| CalculateActivationRangeUint8(activation, shapeOut, |
| &output_activation_min, |
| &output_activation_max); |
| |
| if (needBroadcast) { |
| NNTRACE_COMP_SWITCH("optimized_ops::BroadcastAdd"); |
| #define ANDROID_NN_BROADCAST_ADD(activation) \ |
| tflite::optimized_ops::BroadcastAdd<tflite::FusedActivationFunctionType::activation>( \ |
| left_shift, \ |
| in1, convertShapeToDims(shape1), \ |
| input1_offset, input1_multiplier, input1_shift, \ |
| in2, convertShapeToDims(shape2), \ |
| input2_offset, input2_multiplier, input2_shift, \ |
| output_offset, output_multiplier, output_shift, \ |
| output_activation_min, output_activation_max, \ |
| out, convertShapeToDims(shapeOut)) |
| |
| ANDROID_NN_MACRO_DISPATCH(ANDROID_NN_BROADCAST_ADD) |
| #undef ANDROID_NN_BROADCAST_ADD |
| } else { |
| NNTRACE_COMP_SWITCH("optimized_ops::Add"); |
| #define ANDROID_NN_NORMAL_ADD(activation) \ |
| tflite::optimized_ops::Add<tflite::FusedActivationFunctionType::activation>( \ |
| left_shift, \ |
| in1, convertShapeToDims(shape1), \ |
| input1_offset, input1_multiplier, input1_shift, \ |
| in2, convertShapeToDims(shape2), \ |
| input2_offset, input2_multiplier, input2_shift, \ |
| output_offset, output_multiplier, output_shift, \ |
| output_activation_min, output_activation_max, \ |
| out, convertShapeToDims(shapeOut)) |
| |
| ANDROID_NN_MACRO_DISPATCH(ANDROID_NN_NORMAL_ADD) |
| #undef ANDROID_NN_NORMAL_ADD |
| } |
| |
| return true; |
| } |
| |
| bool mulFloat32(const float* in1, const Shape& shape1, |
| const float* in2, const Shape& shape2, |
| int32_t activation, |
| float* out, const Shape& shapeOut) { |
| NNTRACE_TRANS("mulFloat32"); |
| bool needBroadcast = !SameShape(shape1, shape2); |
| |
| if (needBroadcast) { |
| NNTRACE_COMP_SWITCH("optimized_ops::BroadcastMul"); |
| #define ANDROID_NN_BROADCAST_MUL(activation) \ |
| tflite::optimized_ops::BroadcastMul<tflite::FusedActivationFunctionType::activation>( \ |
| in1, convertShapeToDims(shape1), \ |
| in2, convertShapeToDims(shape2), \ |
| out, convertShapeToDims(shapeOut)) |
| |
| ANDROID_NN_MACRO_DISPATCH(ANDROID_NN_BROADCAST_MUL) |
| #undef ANDROID_NN_BROADCAST_MUL |
| } else { |
| float output_activation_min, output_activation_max; |
| CalculateActivationRangeFloat(activation, &output_activation_min, |
| &output_activation_max); |
| |
| NNTRACE_COMP_SWITCH("optimized_ops::Mul"); |
| tflite::optimized_ops::Mul( |
| in1, convertShapeToDims(shape1), |
| in2, convertShapeToDims(shape2), |
| output_activation_min, output_activation_max, |
| out, convertShapeToDims(shapeOut)); |
| } |
| |
| return true; |
| } |
| |
| bool mulQuant8(const uint8_t* in1, const Shape& shape1, |
| const uint8_t* in2, const Shape& shape2, |
| int32_t activation, |
| uint8_t* out, const Shape& shapeOut) { |
| NNTRACE_TRANS("mulQuant8"); |
| const int32_t input1_offset = -shape1.offset; |
| const int32_t input2_offset = -shape2.offset; |
| const int32_t output_offset = shapeOut.offset; |
| const double input_product_scale = shape1.scale * shape2.scale; |
| const double real_multiplier = input_product_scale / shapeOut.scale; |
| int32 output_multiplier; |
| int output_shift; |
| if (!QuantizeMultiplierSmallerThanOne(real_multiplier, &output_multiplier, |
| &output_shift)) { |
| return false; |
| } |
| int32_t output_activation_min; |
| int32_t output_activation_max; |
| CalculateActivationRangeUint8(activation, shapeOut, |
| &output_activation_min, |
| &output_activation_max); |
| |
| // Use BROADCAST version to handle the normal case. |
| NNTRACE_COMP_SWITCH("optimized_ops::BroadcastMul"); |
| tflite::optimized_ops::BroadcastMul( |
| in1, convertShapeToDims(shape1), input1_offset, |
| in2, convertShapeToDims(shape2), input2_offset, |
| output_offset, output_multiplier, output_shift, |
| output_activation_min, output_activation_max, |
| out, convertShapeToDims(shapeOut)); |
| |
| return true; |
| } |
| |
| bool floorFloat32(const float* inputData, |
| float* outputData, |
| const Shape& shape) { |
| NNTRACE_TRANS("floorFloat32"); |
| tflite::Dims<4> dim = convertShapeToDims(shape); |
| NNTRACE_COMP_SWITCH("optimized_ops::Floor"); |
| tflite::optimized_ops::Floor(inputData, dim, outputData, dim); |
| return true; |
| } |
| |
| bool dequantizeQuant8ToFloat32(const uint8_t* inputData, |
| float* outputData, |
| const Shape& shape) { |
| NNTRACE_TRANS("dequantizeQuant8ToFloat32"); |
| tflite::Dims<4> dim = convertShapeToDims(shape); |
| NNTRACE_COMP_SWITCH("optimized_ops::Dequantize"); |
| tflite::optimized_ops::Dequantize(inputData, dim, |
| shape.offset, shape.scale, |
| outputData, dim); |
| return true; |
| } |
| |
| bool quantizeFloat32ToQuant8(const float* inputData, uint8_t* outputData, |
| const Shape& outputShape) { |
| NNTRACE_COMP("quantizeFloat32ToQuant8"); |
| uint32_t size = getNumberOfElements(outputShape); |
| for (uint32_t i = 0; i < size; ++i) { |
| outputData[i] = static_cast<uint8_t>(std::max<float>( |
| 0, std::min<float>(255, outputShape.offset + |
| std::round(inputData[i] / outputShape.scale)))); |
| } |
| return true; |
| } |
| |
| bool subFloat32(const float* in1, const Shape& shape1, |
| const float* in2, const Shape& shape2, |
| int32_t activation, |
| float* out, const Shape& shapeOut) { |
| NNTRACE_TRANS("subFloat32"); |
| NNTRACE_COMP_SWITCH("optimized_ops::Sub"); |
| tflite::optimized_ops::Sub( |
| in1, convertShapeToDims(shape1), |
| in2, convertShapeToDims(shape2), |
| out, convertShapeToDims(shapeOut)); |
| return true; |
| } |
| |
| bool divFloat32(const float* in1, const Shape& shape1, |
| const float* in2, const Shape& shape2, |
| int32_t activation, |
| float* out, const Shape& shapeOut) { |
| NNTRACE_TRANS("divFloat32"); |
| float output_activation_min, output_activation_max; |
| CalculateActivationRangeFloat(activation, &output_activation_min, |
| &output_activation_max); |
| |
| bool needBroadcast = !SameShape(shape1, shape2); |
| if (needBroadcast) { |
| NNTRACE_COMP_SWITCH("optimized_ops::BroadcastDiv"); |
| tflite::optimized_ops::BroadcastDiv( |
| in1, convertShapeToDims(shape1), |
| in2, convertShapeToDims(shape2), |
| output_activation_min, output_activation_max, |
| out, convertShapeToDims(shapeOut)); |
| } else { |
| NNTRACE_COMP_SWITCH("optimized_ops::Div"); |
| tflite::optimized_ops::Div( |
| in1, convertShapeToDims(shape1), |
| in2, convertShapeToDims(shape2), |
| output_activation_min, output_activation_max, |
| out, convertShapeToDims(shapeOut)); |
| } |
| return true; |
| } |
| |
| bool meanGeneric(const uint8_t* inputData, const Shape& inputShape, |
| const int32_t* axis, const Shape& axisShape, bool keepDims, |
| uint8_t* outputData, const Shape& outputShape) { |
| NNTRACE_TRANS("meanGeneric"); |
| // Creates a temp index to iterate through input data. |
| int32_t* scratchBuffer = new int32_t[getNumberOfDimensions(inputShape)]; |
| |
| // Creates a temp tensor to store resolved axis given input data. |
| int32_t axisSize = static_cast<int32_t>(getSizeOfDimension(axisShape, 0)); |
| int32_t* resolvedAxis = new int32_t[axisSize]; |
| |
| bool result = true; |
| if (inputShape.type == OperandType::TENSOR_FLOAT32) { |
| float* tempSumBuffer = new (std::nothrow) float[getNumberOfElements(outputShape)]; |
| if (!tempSumBuffer) { |
| LOG(ERROR) << "Failed to allocate tempSumBuffer for MEAN"; |
| result = false; |
| } else { |
| NNTRACE_COMP_SWITCH("optimized_ops::Mean"); |
| tflite::reference_ops::Mean<float, float>( |
| const_cast<float*>(reinterpret_cast<const float*>(inputData)), |
| reinterpret_cast<const int*>(inputShape.dimensions.data()), |
| getNumberOfDimensions(inputShape), |
| reinterpret_cast<float*>(outputData), |
| reinterpret_cast<const int*>(outputShape.dimensions.data()), |
| getNumberOfDimensions(outputShape), |
| axis, axisSize, keepDims, scratchBuffer, resolvedAxis, |
| tempSumBuffer); |
| delete[] tempSumBuffer; |
| } |
| } else if (inputShape.type == OperandType::TENSOR_QUANT8_ASYMM) { |
| int32_t* tempSumBuffer = new (std::nothrow) int32_t[getNumberOfElements(outputShape)]; |
| if (!tempSumBuffer) { |
| LOG(ERROR) << "Failed to allocate tempSumBuffer for MEAN"; |
| result = false; |
| } else { |
| NNTRACE_COMP_SWITCH("optimized_ops::Mean"); |
| tflite::reference_ops::Mean<uint8_t, int32_t>( |
| const_cast<uint8_t*>(inputData), |
| reinterpret_cast<const int*>(inputShape.dimensions.data()), |
| getNumberOfDimensions(inputShape), |
| outputData, |
| reinterpret_cast<const int*>(outputShape.dimensions.data()), |
| getNumberOfDimensions(outputShape), |
| axis, axisSize, keepDims, scratchBuffer, resolvedAxis, |
| tempSumBuffer); |
| delete[] tempSumBuffer; |
| } |
| } else { |
| LOG(ERROR) << "Unsupported data type"; |
| result = false; |
| } |
| delete[] scratchBuffer; |
| delete[] resolvedAxis; |
| return result; |
| } |
| |
| bool pReluGeneric(const uint8_t* inputData, const Shape& inputShape, const uint8_t* alphaData, |
| const Shape& alphaShape, uint8_t* outputData, const Shape& outputShape) { |
| NNTRACE_TRANS("pReluGeneric"); |
| if (inputShape.type == OperandType::TENSOR_FLOAT32) { |
| return broadcastOpBase<float>(reinterpret_cast<const float*>(inputData), inputShape, |
| reinterpret_cast<const float*>(alphaData), alphaShape, |
| reinterpret_cast<float*>(outputData), outputShape, |
| [](const float& val1, const float& val2) { |
| return val1 >= 0.0f ? val1 : val1 * val2; |
| }); |
| } else if (inputShape.type == OperandType::TENSOR_QUANT8_ASYMM) { |
| const int32_t input_offset = -inputShape.offset; |
| const int32_t alpha_offset = -alphaShape.offset; |
| const int32_t output_offset = outputShape.offset; |
| const double input_product_scale = inputShape.scale * alphaShape.scale; |
| const double real_multiplier_pos = inputShape.scale / outputShape.scale; |
| const double real_multiplier_neg = input_product_scale / outputShape.scale; |
| int32_t output_multiplier_pos, output_shift_pos; |
| int32_t output_multiplier_neg, output_shift_neg; |
| if (!QuantizeMultiplierSmallerThanOne(real_multiplier_pos, &output_multiplier_pos, |
| &output_shift_pos)) { |
| return false; |
| } |
| if (!QuantizeMultiplierSmallerThanOne(real_multiplier_neg, &output_multiplier_neg, |
| &output_shift_neg)) { |
| return false; |
| } |
| return broadcastOpBase<uint8_t>( |
| reinterpret_cast<const uint8_t*>(inputData), inputShape, |
| reinterpret_cast<const uint8_t*>(alphaData), alphaShape, |
| reinterpret_cast<uint8_t*>(outputData), outputShape, |
| [&](const uint8_t& val1, const uint8_t& val2) { |
| const int32_t input = input_offset + static_cast<int32_t>(val1); |
| int32_t output_val; |
| if (input >= 0) { |
| output_val = output_offset + |
| tflite::MultiplyByQuantizedMultiplierSmallerThanOneExp( |
| input, output_multiplier_pos, -output_shift_pos); |
| } else { |
| const int32_t alpha = alpha_offset + static_cast<int32_t>(val2); |
| output_val = |
| output_offset + |
| tflite::MultiplyByQuantizedMultiplierSmallerThanOneExp( |
| input * alpha, output_multiplier_neg, -output_shift_neg); |
| } |
| return static_cast<uint8_t>(output_val); |
| }); |
| } else { |
| LOG(ERROR) << "Unsupported data type"; |
| return false; |
| } |
| } |
| } // namespace nn |
| } // namespace android |
