Created by: hello-lx

(1).##################################### 环境 ###################################### ubuntu16.04， python=3.6.10

(2).################################################################################ paddle-1.8 GIT COMMIT ID: 1e01335e WITH_MKL: ON WITH_MKLDNN: ON WITH_GPU: ON CUDA version: 10.0 CUDNN version: v7.6 CXX compiler version: 4.8.2

GIT COMMIT ID: 5c2b9258 WITH_MKL: ON WITH_MKLDNN: ON WITH_GPU: ON CUDA version: 10.0 CUDNN version: v7.6 CXX compiler version: 4.8.2

paddle预编译模型：ubuntu14.04_cuda10.0_cudnn7_avx_mkl(两个模型都试了) https://www.paddlepaddle.org.cn/documentation/docs/zh/advanced_guide/inference_deployment/inference/build_and_install_lib_cn.html

(3). ############################# model下载地址 ############################# http://paddle-inference-dist.cdn.bcebos.com/tensorrt_test/mobilenet.tar.gz

(4). ############################# CMakeLists.txt ############################# cmake_minimum_required (VERSION 3.8)

project ("Paddle")

project(cpp_inference_demo CXX C) option(WITH_MKL "Compile demo with MKL/OpenBlas support, default use MKL." ON) option(WITH_GPU "Compile demo with GPU/CPU, default use CPU." ON) option(WITH_STATIC_LIB "Compile demo with static/shared library, default use static." ON) option(USE_TENSORRT "Compile demo with TensorRT." OFF)

macro(safe_set_static_flag) foreach(flag_var CMAKE_CXX_FLAGS CMAKE_CXX_FLAGS_DEBUG CMAKE_CXX_FLAGS_RELEASE CMAKE_CXX_FLAGS_MINSIZEREL CMAKE_CXX_FLAGS_RELWITHDEBINFO) if(${flag_var} MATCHES "/MD") string(REGEX REPLACE "/MD" "/MT" ${flag_var} "${${flag_var}}") endif(${flag_var} MATCHES "/MD") endforeach(flag_var) endmacro()

set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -g") set(CMAKE_STATIC_LIBRARY_PREFIX "") message("flags" ${CMAKE_CXX_FLAGS})

set(PADDLE_LIB "/home/yuan/XSpace/Env-Test/Paddle/fluid_inference") set(DEMO_NAME "mobilenet_test") set(CUDA_LIB "/usr/local/cuda-10.0/lib64") set(CUDNN_LIB "/usr/local/cuda-10.0/lib64") set(CMAKE_SHARED_LIBRARY_SUFFIX ".so")

if(NOT DEFINED PADDLE_LIB) message(FATAL_ERROR "please set PADDLE_LIB with -DPADDLE_LIB=/path/paddle/lib") endif() if(NOT DEFINED DEMO_NAME) message(FATAL_ERROR "please set DEMO_NAME with -DDEMO_NAME=demo_name") endif()

if (USE_TENSORRT AND WITH_GPU) include_directories("${TENSORRT_ROOT}/include") link_directories("${TENSORRT_ROOT}/lib") endif()

include_directories(${PADDLE_LIB}) include_directories(${PADDLE_LIB}/third_party/install/protobuf/include) include_directories(${PADDLE_LIB}/third_party/install/glog/include) include_directories(${PADDLE_LIB}/third_party/install/gflags/include) include_directories(${PADDLE_LIB}/third_party/install/xxhash/include) include_directories(${PADDLE_LIB}/third_party/install/zlib/include) include_directories(${PADDLE_LIB}/third_party/boost) include_directories(${PADDLE_LIB}/third_party/eigen3)

add_executable(${DEMO_NAME} pp_test.cc)

if(WITH_MKL) include_directories("${PADDLE_LIB}/third_party/install/mklml/include") set(MATH_LIB ${PADDLE_LIB}/third_party/install/mklml/lib/libmklml_intel${CMAKE_SHARED_LIBRARY_SUFFIX} ${PADDLE_LIB}/third_party/install/mklml/lib/libiomp5${CMAKE_SHARED_LIBRARY_SUFFIX}) set(MKLDNN_PATH "${PADDLE_LIB}/third_party/install/mkldnn") if(EXISTS ${MKLDNN_PATH}) include_directories("${MKLDNN_PATH}/include") set(MKLDNN_LIB ${MKLDNN_PATH}/lib/libmkldnn.so.0) endif() else() set(MATH_LIB ${PADDLE_LIB}/third_party/install/openblas/lib/libopenblas${CMAKE_STATIC_LIBRARY_SUFFIX}) endif()

# Note: libpaddle_inference_api.so/a must put before libpaddle_fluid.so/a if(WITH_STATIC_LIB) set(DEPS ${PADDLE_LIB}/paddle/lib/libpaddle_fluid${CMAKE_STATIC_LIBRARY_SUFFIX}) else() set(DEPS ${PADDLE_LIB}/paddle/lib/libpaddle_fluid${CMAKE_SHARED_LIBRARY_SUFFIX}) endif()

set(EXTERNAL_LIB "-lrt -ldl -lpthread") set(DEPS ${DEPS} ${MATH_LIB} ${MKLDNN_LIB} glog gflags protobuf z xxhash ${EXTERNAL_LIB})

if(WITH_GPU) # if (USE_TENSORRT) # set(DEPS ${DEPS} # ${TENSORRT_ROOT}/lib/libnvinfer${CMAKE_SHARED_LIBRARY_SUFFIX}) # set(DEPS ${DEPS} # ${TENSORRT_ROOT}/lib/libnvinfer_plugin${CMAKE_SHARED_LIBRARY_SUFFIX}) # endif() set(DEPS ${DEPS} ${CUDA_LIB}/libcudart${CMAKE_SHARED_LIBRARY_SUFFIX} ) set(DEPS ${DEPS} ${CUDA_LIB}/libcudart${CMAKE_SHARED_LIBRARY_SUFFIX} ) set(DEPS ${DEPS} ${CUDA_LIB}/libcublas${CMAKE_SHARED_LIBRARY_SUFFIX} ) set(DEPS ${DEPS} ${CUDNN_LIB}/libcudnn${CMAKE_SHARED_LIBRARY_SUFFIX} ) endif()

target_link_libraries(${DEMO_NAME} ${DEPS})

(5). ############################# 测试代码 ############################# #include <gflags/gflags.h> #include <glog/logging.h> #include #include #include #include #include "paddle/include/paddle_inference_api.h"

namespace paddle { void CreateConfig(AnalysisConfig* config, const std::string& model_dirname) { // 模型从磁盘进行加载 config->SetModel(model_dirname + "/model",
model_dirname + "/params");
// config->SetModel(model_dirname); // 如果模型从内存中加载，可以使用SetModelBuffer接口 // config->SetModelBuffer(prog_buffer, prog_size, params_buffer, params_size); config->EnableUseGpu(100 /设定GPU初始显存池为MB/, 0 /设定GPU ID为0/); //开启GPU预测

/* for cpu config->DisableGpu(); config->EnableMKLDNN(); // 开启MKLDNN加速 config->SetCpuMathLibraryNumThreads(10); */

// 使用ZeroCopyTensor，此处必须设置为false config->SwitchUseFeedFetchOps(false); // 若输入为多个，此处必须设置为true config->SwitchSpecifyInputNames(true); config->SwitchIrDebug(true); // 可视化调试选项，若开启，则会在每个图优化过程后生成dot文件 // config->SwitchIrOptim(false); // 默认为true。如果设置为false，关闭所有优化 // config->EnableMemoryOptim(); // 开启内存/显存复用 }

void RunAnalysis(int batch_size, std::string model_dirname) { // 1. 创建AnalysisConfig AnalysisConfig config; CreateConfig(&config, model_dirname);

// 2. 根据config 创建predictor，并准备输入数据，此处以全0数据为例 auto predictor = CreatePaddlePredictor(config); int channels = 3; int height = 224; int width = 224; float input[batch_size * channels * height * width] = {0};

// 3. 创建输入 // 使用了ZeroCopy接口，可以避免预测中多余的CPU copy，提升预测性能 auto input_names = predictor->GetInputNames(); auto input_t = predictor->GetInputTensor(input_names[0]); input_t->Reshape({batch_size, channels, height, width}); input_t->copy_from_cpu(input);

// 4. 运行预测引擎 CHECK(predictor->ZeroCopyRun());

// 5. 获取输出 std::vector out_data; auto output_names = predictor->GetOutputNames(); auto output_t = predictor->GetOutputTensor(output_names[0]); std::vector output_shape = output_t->shape(); int out_num = std::accumulate(output_shape.begin(), output_shape.end(), 1, std::multiplies());

out_data.resize(out_num); output_t->copy_to_cpu(out_data.data()); } } // namespace paddle

int main() { paddle::RunAnalysis(1, "./mobilenet"); return 0; }

(6). ############################# problem ############################# yuan@Y:~/XSpace/Env-Test/Paddle/build$ cmake .. && make flags -std=c++11 -g -- Configuring done -- Generating done -- Build files have been written to: /home/yuan/XSpace/Env-Test/Paddle/build [ 50%] Linking CXX executable mobilenet_test CMakeFiles/mobilenet_test.dir/pp_test.cc.o：在函数‘paddle::CreateConfig(paddle::AnalysisConfig*, std::__cxx11::basic_string<char, std::char_traits, std::allocator > const&)’中： /home/yuan/XSpace/Env-Test/Paddle/pp_test.cc:14：对‘paddle::AnalysisConfig::SetModel(std::__cxx11::basic_string<char, std::char_traits, std::allocator > const&, std::__cxx11::basic_string<char, std::char_traits, std::allocator > const&)’未定义的引用 collect2: error: ld returned 1 exit status CMakeFiles/mobilenet_test.dir/build.make:91: recipe for target 'mobilenet_test' failed make[2]: *** [mobilenet_test] Error 1 CMakeFiles/Makefile2:72: recipe for target 'CMakeFiles/mobilenet_test.dir/all' failed make[1]: *** [CMakeFiles/mobilenet_test.dir/all] Error 2 Makefile:83: recipe for target 'all' failed make: *** [all] Error 2 yuan@Y:~/XSpace/Env-Test/Paddle/build$