每次启动时间过长，能否将生成的中间文件保存下来？

[MirrorYuChen]

每次启动，tensorrt都会对caffe模型进行转换，能否只转换一次，并将转换后的模型保存到本地，后面每次只读取转换后的模型？

[PKUZHOU]

可以的，tensorRT可以从文件加载模型

[MirrorYuChen]

我按照tensorRT-inference里面的代码将您程序中baseEngine.cpp改成如下形式： `void baseEngine::caffeToGIEModel(const std::string &deployFile, // name for caffe prototxt const std::string &modelFile, // name for model const std::vector &outputs, // network outputs unsigned int maxBatchSize, // batch size - NB must be at least as large as the batch we want to run with) IHostMemory *&gieModelStream) // output buffer for the GIE model {
stringstream gieStream; gieStream.seekg(0, gieStream.beg);

char cache_file[512];
sprintf(cache_file, "%s.engine", modelFile.c_str());
std::ifstream cache(cache_file);
if (!cache) {
    // create the builder
    IBuilder *builder = createInferBuilder(gLogger);

    // parse the caffe model to populate the network, then set the outputs
    INetworkDefinition *network = builder->createNetwork();
    ICaffeParser *parser = createCaffeParser();

    const IBlobNameToTensor *blobNameToTensor = parser->parse(deployFile.c_str(),
                                                            modelFile.c_str(),
                                                            *network,
                                                            nvinfer1::DataType::kHALF);
    // specify which tensors are outputs
    for (auto &s : outputs)
        network->markOutput(*blobNameToTensor->find(s.c_str()));

    // Build the engine
    builder->setMaxBatchSize(maxBatchSize);
    builder->setMaxWorkspaceSize(1 << 25);
    builder->setHalf2Mode(true);

    // cuda engie
    ICudaEngine*engine = builder->buildCudaEngine(*network);
    assert(engine);
    nvinfer1::IHostMemory* serMem = engine->serialize();
    gieStream.write((const char*)serMem->data(), serMem->size());

    // we don't need the network any more, and we can destroy the parser
    network->destroy();
    parser->destroy();
    builder->destroy();

    std::ofstream outFile;
    outFile.open(cache_file);
    outFile << gieStream.rdbuf();
    outFile.close();
    gieStream.seekg(0, gieStream.beg);    
} else {
    printf("loading network profile from engine cache... %s\n", cache_file);
    gieStream << cache.rdbuf();
    cache.close();
}
nvinfer1::IRuntime* infer = nvinfer1::createInferRuntime(gLogger);
gieStream.seekg(0, std::ios::end);
const int modelSize = gieStream.tellg();
gieStream.seekg(0, std::ios::beg);
void* modelMem = malloc(modelSize);
gieStream.read((char*)modelMem, modelSize);
nvinfer1::ICudaEngine* engine = infer->deserializeCudaEngine(modelMem, modelSize, NULL);
free(modelMem);
assert(engine);
context = engine->createExecutionContext();    

}

` 代码能够编译通过，运行也能产生中间文件，但是没有输出结果，大佬有时间能指导一下怎么改一下吗？