init_tengine() 在rv1126板子上npu运行时初始化时间很久

[blueskywwc]

在板子上使用模型进行推理时速度很快，但j加载模型初始化时间很久，大概十几秒，能否有某种预编译模型的方式来把初始化速度降低到1秒以下呢，谢谢

[BUG1989]

CMake 构建工程时，请启用 TENGINE_ENABLE_MODEL_CACHE 功能 -DTENGINE_ENABLE_MODEL_CACHE=ON

[blueskywwc]

谢谢，但是我启用后没啥效果

cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/arm-linux-gnueabihf.toolchain.cmake -DTENGINE_ENABLE_MODEL_CACHE=ON -DTENGINE_ENABLE_TIM_VX=ON -DTENGINE_OPENMP=OFF .. -- The C compiler identification is GNU 8.3.0 -- The CXX compiler identification is GNU 8.3.0 -- Check for working C compiler: /data/Tengine0407/rv1126/rv1126_rv1109_linux_sdk_v1.8.0_20210224/prebuilts/gcc/linux-x86/arm/gcc-arm-8.3-2019.03-x86_64-arm-linux-gnueabihf/bin/arm-linux-gnueabihf-gcc -- Check for working C compiler: /data/Tengine0407/rv1126/rv1126_rv1109_linux_sdk_v1.8.0_20210224/prebuilts/gcc/linux-x86/arm/gcc-arm-8.3-2019.03-x86_64-arm-linux-gnueabihf/bin/arm-linux-gnueabihf-gcc - works -- Detecting C compiler ABI info -- Detecting C compiler ABI info - done -- Detecting C compile features -- Detecting C compile features - done -- Check for working CXX compiler: /data/Tengine0407/rv1126/rv1126_rv1109_linux_sdk_v1.8.0_20210224/prebuilts/gcc/linux-x86/arm/gcc-arm-8.3-2019.03-x86_64-arm-linux-gnueabihf/bin/arm-linux-gnueabihf-g++ -- Check for working CXX compiler: /data/Tengine0407/rv1126/rv1126_rv1109_linux_sdk_v1.8.0_20210224/prebuilts/gcc/linux-x86/arm/gcc-arm-8.3-2019.03-x86_64-arm-linux-gnueabihf/bin/arm-linux-gnueabihf-g++ - works -- Detecting CXX compiler ABI info -- Detecting CXX compiler ABI info - done -- Detecting CXX compile features -- Detecting CXX compile features - done -- Configuring done -- Generating done **CMake Warning: Manually-specified variables were not used by the project:

TENGINE_ENABLE_MODEL_CACHE
TENGINE_ENABLE_TIM_VX
TENGINE_OPENMP**

为啥指定开启模型缓存没被使用呢，谢谢

-- Build files have been written to: /data/code_cross_compile/tengine_code/Tengine_palmvein/build

[blueskywwc]

你好，是我的问题描述不准确，实际是prerun_graph_multithread(graph, opt)执行起来很耗时间，大概23秒，有没有什么方式解决呢，谢谢！