| // Copyright (c) 2023-2024, ARM Limited. |
| // |
| // 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. |
| |
| #include "generate_utils.h" |
| |
| #include <nlohmann/json.hpp> |
| |
| #include <algorithm> |
| |
| namespace tosa |
| { |
| |
| NLOHMANN_JSON_SERIALIZE_ENUM(DType, |
| { |
| { DType::DType_UNKNOWN, "UNKNOWN" }, |
| { DType::DType_BOOL, "BOOL" }, |
| { DType::DType_INT4, "INT4" }, |
| { DType::DType_INT8, "INT8" }, |
| { DType::DType_INT16, "INT16" }, |
| { DType::DType_INT32, "INT32" }, |
| { DType::DType_INT48, "INT48" }, |
| { DType::DType_FP16, "FP16" }, |
| { DType::DType_BF16, "BF16" }, |
| { DType::DType_FP32, "FP32" }, |
| { DType::DType_SHAPE, "SHAPE" }, |
| }) |
| |
| NLOHMANN_JSON_SERIALIZE_ENUM(Op, |
| { |
| { Op::Op_UNKNOWN, "UNKNOWN" }, |
| { Op::Op_ABS, "ABS" }, |
| { Op::Op_ADD, "ADD" }, |
| { Op::Op_ARGMAX, "ARGMAX" }, |
| { Op::Op_AVG_POOL2D, "AVG_POOL2D" }, |
| { Op::Op_CAST, "CAST" }, |
| { Op::Op_CEIL, "CEIL" }, |
| { Op::Op_CLAMP, "CLAMP" }, |
| { Op::Op_CONCAT, "CONCAT" }, |
| { Op::Op_CONV2D, "CONV2D" }, |
| { Op::Op_EQUAL, "EQUAL" }, |
| { Op::Op_ERF, "ERF" }, |
| { Op::Op_EXP, "EXP" }, |
| { Op::Op_FLOOR, "FLOOR" }, |
| { Op::Op_FULLY_CONNECTED, "FULLY_CONNECTED" }, |
| { Op::Op_GATHER, "GATHER" }, |
| { Op::Op_GREATER, "GREATER" }, |
| { Op::Op_GREATER_EQUAL, "GREATER_EQUAL" }, |
| { Op::Op_IDENTITY, "IDENTITY" }, |
| { Op::Op_LOG, "LOG" }, |
| { Op::Op_MATMUL, "MATMUL" }, |
| { Op::Op_MAXIMUM, "MAXIMUM" }, |
| { Op::Op_MAX_POOL2D, "MAX_POOL2D" }, |
| { Op::Op_MINIMUM, "MINIMUM" }, |
| { Op::Op_MUL, "MUL" }, |
| { Op::Op_NEGATE, "NEGATE" }, |
| { Op::Op_PAD, "PAD" }, |
| { Op::Op_POW, "POW" }, |
| { Op::Op_RECIPROCAL, "RECIPROCAL" }, |
| { Op::Op_RESHAPE, "RESHAPE" }, |
| { Op::Op_RSQRT, "RSQRT" }, |
| { Op::Op_REDUCE_MAX, "REDUCE_MAX" }, |
| { Op::Op_REDUCE_MIN, "REDUCE_MIN" }, |
| { Op::Op_REDUCE_PRODUCT, "REDUCE_PRODUCT" }, |
| { Op::Op_REDUCE_SUM, "REDUCE_SUM" }, |
| { Op::Op_SCATTER, "SCATTER" }, |
| { Op::Op_SELECT, "SELECT" }, |
| { Op::Op_SIGMOID, "SIGMOID" }, |
| { Op::Op_SUB, "SUB" }, |
| { Op::Op_TANH, "TANH" }, |
| { Op::Op_TILE, "TILE" }, |
| }) |
| |
| } // namespace tosa |
| |
| namespace TosaReference |
| { |
| |
| NLOHMANN_JSON_SERIALIZE_ENUM(GeneratorType, |
| { |
| { GeneratorType::Unknown, "UNKNOWN" }, |
| { GeneratorType::PseudoRandom, "PSEUDO_RANDOM" }, |
| { GeneratorType::DotProduct, "DOT_PRODUCT" }, |
| { GeneratorType::OpFullRange, "OP_FULL_RANGE" }, |
| { GeneratorType::OpBoundary, "OP_BOUNDARY" }, |
| { GeneratorType::OpSpecial, "OP_SPECIAL" }, |
| { GeneratorType::FixedData, "FIXED_DATA" }, |
| }) |
| |
| // NOTE: This assumes it's VARIABLE if the InputType is not recognized |
| NLOHMANN_JSON_SERIALIZE_ENUM(InputType, |
| { |
| { InputType::Variable, "VARIABLE" }, |
| { InputType::Constant, "CONSTANT" }, |
| }) |
| |
| void from_json(const nlohmann::json& j, DotProductInfo& dotProductInfo) |
| { |
| j.at("s").get_to(dotProductInfo.s); |
| j.at("ks").get_to(dotProductInfo.ks); |
| j.at("acc_type").get_to(dotProductInfo.accType); |
| if (j.contains("kernel")) |
| { |
| j.at("kernel").get_to(dotProductInfo.kernel); |
| } |
| if (j.contains("axis")) |
| { |
| j.at("axis").get_to(dotProductInfo.axis); |
| } |
| } |
| |
| void from_json(const nlohmann::json& j, PseudoRandomInfo& pseudoRandomInfo) |
| { |
| j.at("rng_seed").get_to(pseudoRandomInfo.rngSeed); |
| if (j.contains("range")) |
| { |
| j.at("range").get_to(pseudoRandomInfo.range); |
| } |
| if (j.contains("round")) |
| { |
| j.at("round").get_to(pseudoRandomInfo.round); |
| } |
| } |
| |
| void from_json(const nlohmann::json& j, FixedDataInfo& fixedDataInfo) |
| { |
| j.at("data").get_to(fixedDataInfo.data); |
| } |
| |
| void from_json(const nlohmann::json& j, GenerateConfig& cfg) |
| { |
| j.at("data_type").get_to(cfg.dataType); |
| j.at("input_type").get_to(cfg.inputType); |
| j.at("shape").get_to(cfg.shape); |
| j.at("input_pos").get_to(cfg.inputPos); |
| j.at("op").get_to(cfg.opType); |
| j.at("generator").get_to(cfg.generatorType); |
| |
| // Set up defaults for dotProductInfo |
| cfg.dotProductInfo.s = -1; |
| cfg.dotProductInfo.ks = -1; |
| cfg.dotProductInfo.accType = DType_UNKNOWN; |
| cfg.dotProductInfo.kernel = std::vector<int32_t>(); |
| cfg.dotProductInfo.axis = -1; |
| if (j.contains("dot_product_info")) |
| { |
| j.at("dot_product_info").get_to(cfg.dotProductInfo); |
| } |
| |
| // Set up defaults for pseudoRandomInfo |
| cfg.pseudoRandomInfo.rngSeed = -1; |
| cfg.pseudoRandomInfo.range = std::vector<std::string>(); |
| cfg.pseudoRandomInfo.round = false; |
| if (j.contains("pseudo_random_info")) |
| { |
| j.at("pseudo_random_info").get_to(cfg.pseudoRandomInfo); |
| } |
| |
| // Set up defaults for fixedDataInfo |
| cfg.fixedDataInfo.data = std::vector<int32_t>(); |
| if (j.contains("fixed_data_info")) |
| { |
| j.at("fixed_data_info").get_to(cfg.fixedDataInfo); |
| } |
| } |
| |
| std::optional<GenerateConfig> parseGenerateConfig(const char* json, const char* tensorName) |
| { |
| if (!tensorName) |
| return std::nullopt; |
| |
| auto jsonCfg = nlohmann::json::parse(json, nullptr, /* allow exceptions */ false); |
| |
| if (jsonCfg.is_discarded()) |
| { |
| WARNING("[Generator] Invalid json config."); |
| return std::nullopt; |
| } |
| if (!jsonCfg.contains("tensors")) |
| { |
| WARNING("[Generator] Missing tensors in json config."); |
| return std::nullopt; |
| } |
| const auto& tensors = jsonCfg["tensors"]; |
| if (!tensors.contains(tensorName)) |
| { |
| WARNING("[Generator] Missing tensor %s in json config.", tensorName); |
| return std::nullopt; |
| } |
| const auto& namedTensor = tensors[tensorName]; |
| return namedTensor.get<GenerateConfig>(); |
| } |
| |
| int64_t numElementsFromShape(const std::vector<int32_t>& shape) |
| { |
| // Rank 0 shapes have no entries and so this will return 1 |
| // Other ranked shapes will return the product of their dimensions |
| return std::accumulate(std::begin(shape), std::end(shape), 1, std::multiplies<int64_t>()); |
| } |
| |
| size_t elementSizeFromType(DType type) |
| { |
| switch (type) |
| { |
| case DType::DType_BOOL: |
| case DType::DType_UINT8: |
| case DType::DType_INT8: |
| return 1; |
| case DType::DType_UINT16: |
| case DType::DType_INT16: |
| case DType::DType_FP16: |
| case DType::DType_BF16: |
| return 2; |
| case DType::DType_INT32: |
| case DType::DType_FP32: |
| case DType::DType_SHAPE: |
| return 4; |
| default: |
| return 0; |
| } |
| return 0; |
| } |
| } // namespace TosaReference |