nnrt_utils.cpp 15.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37
/*
 * Copyright (c) 2022 Huawei Device Co., Ltd.
 * 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 "nnrt_utils.h"
#include "const.h"
#include <fstream>

namespace OHOS {
namespace NeuralNetworkRuntime {
namespace Test {

OH_NN_UInt32Array TransformUInt32Array(const std::vector<uint32_t>& vector)
{
    uint32_t* data = (vector.empty()) ? nullptr : const_cast<uint32_t*>(vector.data());
    return {data, vector.size()};
}

int BuildMultiOpGraph(OH_NNModel *model, const OHNNGraphArgsMulti &graphArgs)
{
    int ret = 0;
    int opCnt = 0;
    for (int j = 0; j < graphArgs.operationTypes.size(); j++) {
        for (int i = 0; i < graphArgs.operands[j].size(); i++) {
            const OHNNOperandTest &operandTem = graphArgs.operands[j][i];
            auto quantParam = operandTem.quantParam;
            OH_NN_Tensor operand = {operandTem.dataType, (uint32_t) operandTem.shape.size(),
T
tangshihua 已提交
38
            operandTem.shape.data(), quantParam, operandTem.type};
39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58
            ret = OH_NNModel_AddTensor(model, &operand);
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] OH_NNModel_AddTensor failed! ret=%d\n", ret);
                return ret;
            }
            if (std::find(graphArgs.paramIndices[j].begin(), graphArgs.paramIndices[j].end(), opCnt) !=
                graphArgs.paramIndices[j].end()) {
                ret = OH_NNModel_SetTensorData(model, opCnt, operandTem.data, operandTem.length);
            }
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] OH_NNModel_SetTensorData failed! ret=%d\n", ret);
                return ret;
            }
            opCnt += 1;
        }
        auto paramIndices = TransformUInt32Array(graphArgs.paramIndices[j]);
        auto inputIndices = TransformUInt32Array(graphArgs.inputIndices[j]);
        auto outputIndices = TransformUInt32Array(graphArgs.outputIndices[j]);

        ret = OH_NNModel_AddOperation(model, graphArgs.operationTypes[j], &paramIndices, &inputIndices,
T
tangshihua 已提交
59
        &outputIndices);
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNModel_AddOperation failed! ret=%d\n", ret);
            return ret;
        }
    }
    auto graphInputs = TransformUInt32Array(graphArgs.graphInput);
    auto graphOutputs = TransformUInt32Array(graphArgs.graphOutput);
    ret = OH_NNModel_SpecifyInputsAndOutputs(model, &graphInputs, &graphOutputs);
    if (ret != OH_NN_SUCCESS) {
        LOGE("[NNRtTest] OH_NNModel_SpecifyInputsAndOutputs failed! ret=%d\n", ret);
        return ret;
    }
    ret = OH_NNModel_Finish(model);
    if (ret != OH_NN_SUCCESS) {
        LOGE("[NNRtTest] OH_NNModel_Finish failed! ret=%d\n", ret);
        return ret;
    }
    return ret;
}

int BuildSingleOpGraph(OH_NNModel *model, const OHNNGraphArgs &graphArgs)
{
    int ret = 0;
    for (int i = 0; i < graphArgs.operands.size(); i++) {
        const OHNNOperandTest &operandTem = graphArgs.operands[i];
        auto quantParam = operandTem.quantParam;
        OH_NN_Tensor operand = {operandTem.dataType, (uint32_t) operandTem.shape.size(),
T
tangshihua 已提交
87
        operandTem.shape.data(), quantParam, operandTem.type};
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
        ret = OH_NNModel_AddTensor(model, &operand);
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNModel_AddTensor failed! ret=%d\n", ret);
            return ret;
        }

        if (std::find(graphArgs.paramIndices.begin(), graphArgs.paramIndices.end(), i) !=
            graphArgs.paramIndices.end()) {
            ret = OH_NNModel_SetTensorData(model, i, operandTem.data, operandTem.length);
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] OH_NNModel_SetTensorData failed! ret=%d\n", ret);
                return ret;
            }
        }
    }
    auto paramIndices = TransformUInt32Array(graphArgs.paramIndices);
    auto inputIndices = TransformUInt32Array(graphArgs.inputIndices);
    auto outputIndices = TransformUInt32Array(graphArgs.outputIndices);
    if (graphArgs.addOperation) {
        ret = OH_NNModel_AddOperation(model, graphArgs.operationType, &paramIndices, &inputIndices,
                                      &outputIndices);
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNModel_AddOperation failed! ret=%d\n", ret);
            return ret;
        }
    }
    if (graphArgs.specifyIO) {
        ret = OH_NNModel_SpecifyInputsAndOutputs(model, &inputIndices, &outputIndices);
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNModel_SpecifyInputsAndOutputs failed! ret=%d\n", ret);
            return ret;
        }
    }
    if (graphArgs.build) {
        ret = OH_NNModel_Finish(model);
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNModel_Finish failed! ret=%d\n", ret);
            return ret;
        }
    }
    return ret;
}

int SetDevice(OH_NNCompilation *compilation)
{
    int ret = 0;
    const size_t *devicesID{nullptr};
    uint32_t devicesCount{0};
    ret = OH_NNDevice_GetAllDevicesID(&devicesID, &devicesCount);
    if (ret != OH_NN_SUCCESS) {
        LOGE("[NNRtTest] OH_NNDevice_GetAllDevicesID failed! ret=%d\n", ret);
        return ret;
    }
    if (devicesCount <= NO_DEVICE_COUNT) {
        return OH_NN_FAILED;
    }
    size_t targetDevice = devicesID[0]; // Use the first device in system test.
    ret = OH_NNCompilation_SetDevice(compilation, targetDevice);
    return ret;
}

int CompileGraphMock(OH_NNCompilation *compilation, const OHNNCompileParam &compileParam)
{
    int ret = 0;
    ret = SetDevice(compilation);
    if (ret != OH_NN_SUCCESS) {
        LOGE("[NNRtTest] OH_NNCompilation_SetDevice failed! ret=%d\n", ret);
        return ret;
    }
    // set cache
    if (!compileParam.cacheDir.empty()) {
        ret = OH_NNCompilation_SetCache(compilation, compileParam.cacheDir.c_str(),
T
tangshihua 已提交
160
        compileParam.cacheVersion);
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNCompilation_SetCache failed! ret=%d\n", ret);
            return ret;
        }
    }
    // set performance
    if (compileParam.performanceMode != OH_NN_PERFORMANCE_NONE) {
        ret = OH_NNCompilation_SetPerformanceMode(compilation, compileParam.performanceMode);
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNCompilation_SetPerformanceMode failed! ret=%d\n", ret);
            return ret;
        }
    }
    // set priority
    if (compileParam.priority != OH_NN_PRIORITY_NONE) {
        ret = OH_NNCompilation_SetPriority(compilation, compileParam.priority);
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNCompilation_SetPriority failed! ret=%d\n", ret);
            return ret;
        }
    }
    // enable fp16
    if (compileParam.enableFp16) {
        ret = OH_NNCompilation_EnableFloat16(compilation, compileParam.enableFp16);
        if (ret != OH_NN_SUCCESS) {
            LOGE("[NNRtTest] OH_NNCompilation_EnableFloat16 failed! ret=%d\n", ret);
            return ret;
        }
    }
    // build
    ret = OH_NNCompilation_Build(compilation);
    return ret;
}


int ExecuteGraphMock(OH_NNExecutor *executor, const OHNNGraphArgs &graphArgs,
T
tangshihua 已提交
197
float* expect)
198 199 200 201 202 203 204 205 206
{
    OHOS::sptr<V1_0::MockIDevice> device = V1_0::MockIDevice::GetInstance();
    int ret = 0;
    uint32_t inputIndex = 0;
    uint32_t outputIndex = 0;
    for (auto i = 0; i < graphArgs.operands.size(); i++) {
        const OHNNOperandTest &operandTem = graphArgs.operands[i];
        auto quantParam = operandTem.quantParam;
        OH_NN_Tensor operand = {operandTem.dataType, (uint32_t) operandTem.shape.size(),
T
tangshihua 已提交
207 208
        operandTem.shape.data(), 
        quantParam, operandTem.type};
209 210 211
        if (std::find(graphArgs.inputIndices.begin(), graphArgs.inputIndices.end(), i) !=
            graphArgs.inputIndices.end()) {
            ret = OH_NNExecutor_SetInput(executor, inputIndex, &operand, operandTem.data,
T
tangshihua 已提交
212
            operandTem.length);
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] OH_NNExecutor_SetInput failed! ret=%d\n", ret);
                return ret;
            }
            inputIndex += 1;
        } else if (std::find(graphArgs.outputIndices.begin(), graphArgs.outputIndices.end(), i) !=
                   graphArgs.outputIndices.end()) {
            ret = OH_NNExecutor_SetOutput(executor, outputIndex, operandTem.data, operandTem.length);
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] OH_NNExecutor_SetOutput failed! ret=%d\n", ret);
                return ret;
            }
            ret = device->MemoryCopy(expect, operandTem.length);
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] device set expect output failed! ret=%d\n", ret);
                return ret;
            }
            outputIndex += 1;
        }
    }
    ret = OH_NNExecutor_Run(executor);
    return ret;
}

T
tangshihua 已提交
237
int ExecutorWithMemory(OH_NNExecutor *executor, const OHNNGraphArgs &graphArgs, OH_NN_Memory *OHNNMemory[], float *expect)
238 239 240 241 242 243 244 245 246
{
    OHOS::sptr<V1_0::MockIDevice> device = V1_0::MockIDevice::GetInstance();
    int ret = 0;
    uint32_t inputIndex = 0;
    uint32_t outputIndex = 0;
    for (auto i = 0; i < graphArgs.operands.size(); i++) {
        const OHNNOperandTest &operandTem = graphArgs.operands[i];
        auto quantParam = operandTem.quantParam;
        OH_NN_Tensor operand = {operandTem.dataType, (uint32_t) operandTem.shape.size(),
T
tangshihua 已提交
247 248
        operandTem.shape.data(),
        quantParam, operandTem.type};
249 250 251
        if (std::find(graphArgs.inputIndices.begin(), graphArgs.inputIndices.end(), i) !=
            graphArgs.inputIndices.end()) {
            OH_NN_Memory *inputMemory = OH_NNExecutor_AllocateInputMemory(executor, inputIndex,
T
tangshihua 已提交
252
            operandTem.length);
253 254 255 256 257 258 259 260 261 262 263
            ret = OH_NNExecutor_SetInputWithMemory(executor, inputIndex, &operand, inputMemory);
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] OH_NNExecutor_SetInputWithMemory failed! ret=%d\n", ret);
                return ret;
            }
            memcpy_s(inputMemory->data, operandTem.length, (void *) operandTem.data, operandTem.length);
            OHNNMemory[inputIndex] = inputMemory;
            inputIndex += 1;
        } else if (std::find(graphArgs.outputIndices.begin(), graphArgs.outputIndices.end(), i) !=
                   graphArgs.outputIndices.end()) {
            OH_NN_Memory *outputMemory = OH_NNExecutor_AllocateOutputMemory(executor, outputIndex,
T
tangshihua 已提交
264
            operandTem.length);                                                           
265 266 267 268 269 270 271 272 273
            ret = OH_NNExecutor_SetOutputWithMemory(executor, outputIndex, outputMemory);
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] OH_NNExecutor_SetOutputWithMemory failed! ret=%d\n", ret);
                return ret;
            }
            ret = device->MemoryCopy(expect, operandTem.length);
            if (ret != OH_NN_SUCCESS) {
                LOGE("[NNRtTest] device set expect output failed! ret=%d\n", ret);
                return ret;
T
tangshihua 已提交
274
            }
275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
            OHNNMemory[inputIndex + outputIndex] = outputMemory;
            outputIndex += 1;
        }
    }
    ret = OH_NNExecutor_Run(executor);
    return ret;
}


void Free(OH_NNModel *model, OH_NNCompilation *compilation, OH_NNExecutor *executor)
{
    if (model != nullptr) {
        OH_NNModel_Destroy(&model);
        ASSERT_EQ(nullptr, model);
    }
    if (compilation != nullptr) {
        OH_NNCompilation_Destroy(&compilation);
        ASSERT_EQ(nullptr, compilation);
    }
    if (executor != nullptr) {
        OH_NNExecutor_Destroy(&executor);
        ASSERT_EQ(nullptr, executor);
    }
}

PathType CheckPath(const std::string &path)
{
    if (path.empty()) {
        LOGI("CheckPath: path is null");
        return PathType::NOT_FOUND;
    }
    struct stat buf{};
    if (stat(path.c_str(), &buf) == 0) {
        if (buf.st_mode & S_IFDIR) {
            return PathType::DIR;
        } else if (buf.st_mode & S_IFREG) {
            return PathType::FILE;
        } else {
            return PathType::UNKNOWN;
        }
    }
    LOGI("%s not found", path.c_str());
    return PathType::NOT_FOUND;
}

bool DeleteFile(const std::string &path)
{
    if (path.empty()) {
        LOGI("DeleteFile: path is null");
        return false;
    }
    if (CheckPath(path) == PathType::NOT_FOUND) {
        LOGI("not found: %s", path.c_str());
        return true;
    }
    if (remove(path.c_str()) == 0) {
        LOGI("deleted: %s", path.c_str());
        return true;
    }
    LOGI("delete failed: %s", path.c_str());
    return false;
}

void CopyFile(const std::string &srcPath, const std::string &dstPath)
{
    std::ifstream src(srcPath, std::ios::binary);
    std::ofstream dst(dstPath, std::ios::binary);

    dst << src.rdbuf();
}

std::string ConcatPath(const std::string &str1, const std::string &str2)
{
    // boundary
    if (str2.empty()) {
        return str1;
    }
    if (str1.empty()) {
        return str2;
    }
    // concat
    char end = str1[str1.size() - 1];
    if (end == '\\' or end == '/') {
        return str1 + str2;
    } else {
        return str1 + '/' + str2;
    }
}

void DeleteFolder(const std::string &path)
{
    if (path.empty()) {
        LOGI("DeletePath: path is null");
        return;
    }

    DIR *dir = opendir(path.c_str());
    // check is dir ?
    if (dir == nullptr) {
        LOGE("[NNRtTest] Can not open dir. Check path or permission! path: %s", path.c_str());
        return;
    }
    struct dirent *file;
    // read all the files in dir
    std::vector <std::string> pathList;
    while ((file = readdir(dir)) != nullptr) {
        // skip "." and ".."
        if (strcmp(file->d_name, ".") == 0 || strcmp(file->d_name, "..") == 0) {
            continue;
        }
        if (file->d_type == DT_DIR) {
            std::string filePath = path + "/" + file->d_name;
            DeleteFolder(filePath); // 递归执行
        } else {
            pathList.emplace_back(ConcatPath(path, file->d_name));
        }
    }
    closedir(dir);
    pathList.emplace_back(path);
    LOGI("[Common] Delete folder %s", path.c_str());
    for (auto &i : pathList) {
        DeleteFile(i);
    }
}

bool CreateFolder(const std::string &path)
{
    if (path.empty()) {
        LOGI("CreateFolder: path is empty");
        return false;
    }
    LOGI("CreateFolder:%s", path.c_str());
T
tangshihua 已提交
407
    mode_t mode = 0700;
408 409 410 411 412 413 414 415 416 417
    for (int i = 1; i < path.size() - 1; i++) {
        if (path[i] != '/') {
            continue;
        }
        PathType ret = CheckPath(path.substr(0, i));
        switch (ret) {
            case PathType::DIR:
                continue;
            case PathType::NOT_FOUND:
                LOGI("mkdir: %s", path.substr(0, i).c_str());
T
tangshihua 已提交
418
                mkdir(path.substr(0, i).c_str(), mode);
419 420 421 422 423 424
                break;
            default:
                LOGI("error: %s", path.substr(0, i).c_str());
                return false;
        }
    }
T
tangshihua 已提交
425
    mkdir(path.c_str(), mode);
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448
    return CheckPath(path) == PathType::DIR;
}

bool CheckOutput(const float* output, const float* expect)
{
    if (output == nullptr || expect == nullptr) {
        LOGE("[NNRtTest] output or expect is nullptr\n");
        return false;
    }
    for (int i = 0; i < ELEMENT_COUNT; i++) {
        if (std::abs(float(output[i]) - float(expect[i])) > 1e-8) {
            for (int j = 0; j < ELEMENT_COUNT; j++) {
                LOGE("[NNRtTest] output %d not match: expect:%f, actual:%f\n", j, float(expect[j]), float(output[j]));
            }
            return false;
        }
    }
    return true;
}

} // namespace Test
} // namespace NeuralNetworkRuntime
} // namespace OHOS