1 /* 2 * Copyright (c) 2020-2021 Arm Limited. 3 * 4 * SPDX-License-Identifier: MIT 5 * 6 * Permission is hereby granted, free of charge, to any person obtaining a copy 7 * of this software and associated documentation files (the "Software"), to 8 * deal in the Software without restriction, including without limitation the 9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or 10 * sell copies of the Software, and to permit persons to whom the Software is 11 * furnished to do so, subject to the following conditions: 12 * 13 * The above copyright notice and this permission notice shall be included in all 14 * copies or substantial portions of the Software. 15 * 16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 22 * SOFTWARE. 23 */ 24 #ifndef ARM_COMPUTE_TEST_QLSTM_LAYER_NORMALIZATION_FIXTURE 25 #define ARM_COMPUTE_TEST_QLSTM_LAYER_NORMALIZATION_FIXTURE 26 27 #include "arm_compute/core/TensorShape.h" 28 #include "arm_compute/core/Types.h" 29 #include "tests/AssetsLibrary.h" 30 #include "tests/Globals.h" 31 #include "tests/IAccessor.h" 32 #include "tests/framework/Asserts.h" 33 #include "tests/framework/Fixture.h" 34 #include "tests/validation/Helpers.h" 35 #include "tests/validation/reference/QLSTMLayerNormalization.h" 36 37 namespace arm_compute 38 { 39 namespace test 40 { 41 namespace validation 42 { 43 template <typename TensorType, typename AccessorType, typename FunctionType, typename T> 44 class QLSTMLayerNormalizationValidationFixture : public framework::Fixture 45 { 46 public: 47 template <typename...> setup(TensorShape input_shape,TensorShape weight_shape,TensorShape bias_shape,DataType data_type,QuantizationInfo weight_qinfo)48 void setup(TensorShape input_shape, TensorShape weight_shape, TensorShape bias_shape, DataType data_type, QuantizationInfo weight_qinfo) 49 { 50 ARM_COMPUTE_ERROR_ON(data_type != DataType::QSYMM16); 51 52 _data_type = data_type; 53 _qinfo = weight_qinfo; 54 55 _target = compute_target(input_shape, weight_shape, bias_shape); 56 _reference = compute_reference(input_shape, weight_shape, bias_shape); 57 } 58 59 protected: 60 template <typename InputType, typename BiasType> fill(InputType && input_tensor,InputType && weight_tensor,BiasType && bias_tensor)61 void fill(InputType &&input_tensor, InputType &&weight_tensor, BiasType &&bias_tensor) 62 { 63 switch(_data_type) 64 { 65 case DataType::QSYMM16: 66 { 67 // Value ranges are based on reference implementation's test case. 68 constexpr int16_t input_min = -1000; 69 constexpr int16_t input_max = 1000; 70 constexpr int16_t weight_min = 19000; 71 constexpr int16_t weight_max = 27000; 72 constexpr int32_t bias_min = -16000000; 73 constexpr int32_t bias_max = -13000000; 74 75 std::uniform_int_distribution<> input_distribution(input_min, input_max); 76 std::uniform_int_distribution<> weight_distribution(weight_min, weight_max); 77 std::uniform_int_distribution<> bias_distribution(bias_min, bias_max); 78 79 library->fill(input_tensor, input_distribution, 0); 80 library->fill(weight_tensor, weight_distribution, 0); 81 library->fill(bias_tensor, bias_distribution, 0); 82 break; 83 } 84 default: 85 ARM_COMPUTE_ERROR("non-supported data type"); 86 break; 87 } 88 } 89 allocate_tensors(const std::vector<TensorType * > & tensors)90 void allocate_tensors(const std::vector<TensorType *> &tensors) 91 { 92 for(auto t : tensors) 93 { 94 ARM_COMPUTE_ASSERT(t->info()->is_resizable()); 95 t->allocator()->allocate(); 96 ARM_COMPUTE_ASSERT(!t->info()->is_resizable()); 97 } 98 } 99 compute_target(const TensorShape & input_shape,const TensorShape & weight_shape,const TensorShape & bias_shape)100 TensorType compute_target(const TensorShape &input_shape, const TensorShape &weight_shape, const TensorShape &bias_shape) 101 { 102 TensorType input = create_tensor<TensorType>(input_shape, _data_type, 1); 103 TensorType weight = create_tensor<TensorType>(weight_shape, _data_type, 1, _qinfo); 104 TensorType bias = create_tensor<TensorType>(bias_shape, DataType::S32, 1); 105 TensorType output = create_tensor<TensorType>(input_shape, _data_type, 1); 106 107 FunctionType fn; 108 fn.configure(&input, &output, &weight, &bias); 109 allocate_tensors({ &input, &weight, &bias, &output }); 110 fill(AccessorType(input), AccessorType(weight), AccessorType(bias)); 111 fn.run(); 112 113 return output; 114 } 115 compute_reference(const TensorShape & input_shape,const TensorShape & weight_shape,const TensorShape & bias_shape)116 SimpleTensor<T> compute_reference(const TensorShape &input_shape, const TensorShape &weight_shape, const TensorShape &bias_shape) 117 { 118 // Create reference 119 SimpleTensor<T> input{ input_shape, _data_type, 1 }; 120 SimpleTensor<T> weight{ weight_shape, _data_type, 1, _qinfo }; 121 SimpleTensor<int32_t> bias{ bias_shape, DataType::S32, 1 }; 122 123 // Fill reference 124 fill(input, weight, bias); 125 126 return reference::qlstm_layer_normalization(input, weight, bias); 127 } 128 129 TensorType _target{}; 130 SimpleTensor<T> _reference{}; 131 DataType _data_type{}; 132 QuantizationInfo _qinfo{}; 133 }; 134 } // namespace validation 135 } // namespace test 136 } // namespace arm_compute 137 138 #endif /* ARM_COMPUTE_TEST_QLSTM_LAYER_NORMALIZATION_FIXTURE */ 139