1 // 2 // Copyright © 2020, 2023 Arm Ltd and Contributors. All rights reserved. 3 // SPDX-License-Identifier: MIT 4 // 5 6 #pragma once 7 8 #include "TestUtils.hpp" 9 10 #include <armnn_delegate.hpp> 11 #include <DelegateTestInterpreter.hpp> 12 13 #include <flatbuffers/flatbuffers.h> 14 #include <tensorflow/lite/kernels/register.h> 15 #include <tensorflow/lite/version.h> 16 17 #include <schema_generated.h> 18 19 #include <doctest/doctest.h> 20 21 namespace 22 { 23 CreateLogicalBinaryTfLiteModel(tflite::BuiltinOperator logicalOperatorCode,tflite::TensorType tensorType,const std::vector<int32_t> & input0TensorShape,const std::vector<int32_t> & input1TensorShape,const std::vector<int32_t> & outputTensorShape,float quantScale=1.0f,int quantOffset=0)24 std::vector<char> CreateLogicalBinaryTfLiteModel(tflite::BuiltinOperator logicalOperatorCode, 25 tflite::TensorType tensorType, 26 const std::vector <int32_t>& input0TensorShape, 27 const std::vector <int32_t>& input1TensorShape, 28 const std::vector <int32_t>& outputTensorShape, 29 float quantScale = 1.0f, 30 int quantOffset = 0) 31 { 32 using namespace tflite; 33 flatbuffers::FlatBufferBuilder flatBufferBuilder; 34 35 std::vector<flatbuffers::Offset<tflite::Buffer>> buffers; 36 buffers.push_back(CreateBuffer(flatBufferBuilder)); 37 buffers.push_back(CreateBuffer(flatBufferBuilder)); 38 buffers.push_back(CreateBuffer(flatBufferBuilder)); 39 buffers.push_back(CreateBuffer(flatBufferBuilder)); 40 41 auto quantizationParameters = 42 CreateQuantizationParameters(flatBufferBuilder, 43 0, 44 0, 45 flatBufferBuilder.CreateVector<float>({ quantScale }), 46 flatBufferBuilder.CreateVector<int64_t>({ quantOffset })); 47 48 49 std::array<flatbuffers::Offset<Tensor>, 3> tensors; 50 tensors[0] = CreateTensor(flatBufferBuilder, 51 flatBufferBuilder.CreateVector<int32_t>(input0TensorShape.data(), 52 input0TensorShape.size()), 53 tensorType, 54 1, 55 flatBufferBuilder.CreateString("input_0"), 56 quantizationParameters); 57 tensors[1] = CreateTensor(flatBufferBuilder, 58 flatBufferBuilder.CreateVector<int32_t>(input1TensorShape.data(), 59 input1TensorShape.size()), 60 tensorType, 61 2, 62 flatBufferBuilder.CreateString("input_1"), 63 quantizationParameters); 64 tensors[2] = CreateTensor(flatBufferBuilder, 65 flatBufferBuilder.CreateVector<int32_t>(outputTensorShape.data(), 66 outputTensorShape.size()), 67 tensorType, 68 3, 69 flatBufferBuilder.CreateString("output"), 70 quantizationParameters); 71 72 // create operator 73 tflite::BuiltinOptions operatorBuiltinOptionsType = tflite::BuiltinOptions_NONE; 74 flatbuffers::Offset<void> operatorBuiltinOptions = 0; 75 switch (logicalOperatorCode) 76 { 77 case BuiltinOperator_LOGICAL_AND: 78 { 79 operatorBuiltinOptionsType = BuiltinOptions_LogicalAndOptions; 80 operatorBuiltinOptions = CreateLogicalAndOptions(flatBufferBuilder).Union(); 81 break; 82 } 83 case BuiltinOperator_LOGICAL_OR: 84 { 85 operatorBuiltinOptionsType = BuiltinOptions_LogicalOrOptions; 86 operatorBuiltinOptions = CreateLogicalOrOptions(flatBufferBuilder).Union(); 87 break; 88 } 89 default: 90 break; 91 } 92 const std::vector<int32_t> operatorInputs{ {0, 1} }; 93 const std::vector<int32_t> operatorOutputs{ 2 }; 94 flatbuffers::Offset <Operator> logicalBinaryOperator = 95 CreateOperator(flatBufferBuilder, 96 0, 97 flatBufferBuilder.CreateVector<int32_t>(operatorInputs.data(), operatorInputs.size()), 98 flatBufferBuilder.CreateVector<int32_t>(operatorOutputs.data(), operatorOutputs.size()), 99 operatorBuiltinOptionsType, 100 operatorBuiltinOptions); 101 102 const std::vector<int> subgraphInputs{ {0, 1} }; 103 const std::vector<int> subgraphOutputs{ 2 }; 104 flatbuffers::Offset <SubGraph> subgraph = 105 CreateSubGraph(flatBufferBuilder, 106 flatBufferBuilder.CreateVector(tensors.data(), tensors.size()), 107 flatBufferBuilder.CreateVector<int32_t>(subgraphInputs.data(), subgraphInputs.size()), 108 flatBufferBuilder.CreateVector<int32_t>(subgraphOutputs.data(), subgraphOutputs.size()), 109 flatBufferBuilder.CreateVector(&logicalBinaryOperator, 1)); 110 111 flatbuffers::Offset <flatbuffers::String> modelDescription = 112 flatBufferBuilder.CreateString("ArmnnDelegate: Logical Binary Operator Model"); 113 flatbuffers::Offset <OperatorCode> operatorCode = CreateOperatorCode(flatBufferBuilder, logicalOperatorCode); 114 115 flatbuffers::Offset <Model> flatbufferModel = 116 CreateModel(flatBufferBuilder, 117 TFLITE_SCHEMA_VERSION, 118 flatBufferBuilder.CreateVector(&operatorCode, 1), 119 flatBufferBuilder.CreateVector(&subgraph, 1), 120 modelDescription, 121 flatBufferBuilder.CreateVector(buffers.data(), buffers.size())); 122 123 flatBufferBuilder.Finish(flatbufferModel, armnnDelegate::FILE_IDENTIFIER); 124 125 return std::vector<char>(flatBufferBuilder.GetBufferPointer(), 126 flatBufferBuilder.GetBufferPointer() + flatBufferBuilder.GetSize()); 127 } 128 LogicalBinaryTest(tflite::BuiltinOperator logicalOperatorCode,tflite::TensorType tensorType,std::vector<armnn::BackendId> & backends,std::vector<int32_t> & input0Shape,std::vector<int32_t> & input1Shape,std::vector<int32_t> & expectedOutputShape,std::vector<bool> & input0Values,std::vector<bool> & input1Values,std::vector<bool> & expectedOutputValues,float quantScale=1.0f,int quantOffset=0)129 void LogicalBinaryTest(tflite::BuiltinOperator logicalOperatorCode, 130 tflite::TensorType tensorType, 131 std::vector<armnn::BackendId>& backends, 132 std::vector<int32_t>& input0Shape, 133 std::vector<int32_t>& input1Shape, 134 std::vector<int32_t>& expectedOutputShape, 135 std::vector<bool>& input0Values, 136 std::vector<bool>& input1Values, 137 std::vector<bool>& expectedOutputValues, 138 float quantScale = 1.0f, 139 int quantOffset = 0) 140 { 141 using namespace delegateTestInterpreter; 142 std::vector<char> modelBuffer = CreateLogicalBinaryTfLiteModel(logicalOperatorCode, 143 tensorType, 144 input0Shape, 145 input1Shape, 146 expectedOutputShape, 147 quantScale, 148 quantOffset); 149 150 // Setup interpreter with just TFLite Runtime. 151 auto tfLiteInterpreter = DelegateTestInterpreter(modelBuffer); 152 CHECK(tfLiteInterpreter.AllocateTensors() == kTfLiteOk); 153 CHECK(tfLiteInterpreter.FillInputTensor(input0Values, 0) == kTfLiteOk); 154 CHECK(tfLiteInterpreter.FillInputTensor(input1Values, 1) == kTfLiteOk); 155 CHECK(tfLiteInterpreter.Invoke() == kTfLiteOk); 156 std::vector<bool> tfLiteOutputValues = tfLiteInterpreter.GetOutputResult(0); 157 std::vector<int32_t> tfLiteOutputShape = tfLiteInterpreter.GetOutputShape(0); 158 159 // Setup interpreter with Arm NN Delegate applied. 160 auto armnnInterpreter = DelegateTestInterpreter(modelBuffer, backends); 161 CHECK(armnnInterpreter.AllocateTensors() == kTfLiteOk); 162 CHECK(armnnInterpreter.FillInputTensor(input0Values, 0) == kTfLiteOk); 163 CHECK(armnnInterpreter.FillInputTensor(input1Values, 1) == kTfLiteOk); 164 CHECK(armnnInterpreter.Invoke() == kTfLiteOk); 165 std::vector<bool> armnnOutputValues = armnnInterpreter.GetOutputResult(0); 166 std::vector<int32_t> armnnOutputShape = armnnInterpreter.GetOutputShape(0); 167 168 armnnDelegate::CompareOutputShape(tfLiteOutputShape, armnnOutputShape, expectedOutputShape); 169 170 armnnDelegate::CompareData(expectedOutputValues, armnnOutputValues, expectedOutputValues.size()); 171 armnnDelegate::CompareData(expectedOutputValues, tfLiteOutputValues, expectedOutputValues.size()); 172 armnnDelegate::CompareData(tfLiteOutputValues, armnnOutputValues, expectedOutputValues.size()); 173 174 tfLiteInterpreter.Cleanup(); 175 armnnInterpreter.Cleanup(); 176 } 177 178 } // anonymous namespace