大兄弟，可以更新ncnn的版本了，最新版本测试了一下，速度翻了一倍，实时检测毫无压力了...

[abc1225]

[rivercn]

Build command failed. Error while executing '/home/river/Android/Sdk/cmake/3.6.4111459/bin/cmake' with arguments {--build /home/river/AndroidStudioProjects/mtcnn_AS_1/app/.externalNativeBuild/cmake/debug/armeabi-v7a --target mtcnn} [1/3] Building CXX object CMakeFiles/mtcnn.dir/src/main/cpp/mtcnn.cpp.o [2/3] Building CXX object CMakeFiles/mtcnn.dir/src/main/cpp/mtcnn_jni.cpp.o [3/3] Linking CXX shared library ../../../../build/intermediates/cmake/debug/obj/armeabi-v7a/libmtcnn.so FAILED: : && /home/river/Android/Sdk/ndk-bundle/toolchains/arm-linux-androideabi-4.9/prebuilt/linux-x86_64/bin/arm-linux-androideabi-g++ --sysroot=/home/river/Android/Sdk/ndk-bundle/sysroot -fPIC -isystem /home/river/Android/Sdk/ndk-bundle/sysroot/usr/include/arm-linux-androideabi -D__ANDROID_API__=17 -g -DANDROID -ffunction-sections -funwind-tables -fstack-protector-strong -no-canonical-prefixes -march=armv7-a -mfloat-abi=softfp -mfpu=vfpv3-d16 -mthumb -Wa,--noexecstack -Wformat -Werror=format-security -fopenmp -O2 -fvisibility=hidden -fvisibility-inlines-hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -std=c++11 -frtti -fexceptions -O0 -Wl,--exclude-libs,libgcc.a -Wl,--exclude-libs,libatomic.a --sysroot /home/river/Android/Sdk/ndk-bundle/platforms/android-17/arch-arm -Wl,--build-id -Wl,--warn-shared-textrel -Wl,--fatal-warnings -Wl,--fix-cortex-a8 -Wl,--no-undefined -Wl,-z,noexecstack -Wl,-z,relro -Wl,-z,now -shared -Wl,-soname,libmtcnn.so -o ../../../../build/intermediates/cmake/debug/obj/armeabi-v7a/libmtcnn.so CMakeFiles/mtcnn.dir/src/main/cpp/mtcnn.cpp.o CMakeFiles/mtcnn.dir/src/main/cpp/mtcnn_jni.cpp.o ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a -ljnigraphics -lz /home/river/Android/Sdk/ndk-bundle/platforms/android-17/arch-arm/usr/lib/liblog.so -latomic -lm "/home/river/Android/Sdk/ndk-bundle/sources/cxx-stl/gnu-libstdc++/4.9/libs/armeabi-v7a/libgnustl_static.a" && : ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function ncnn::Mat::substract_mean_normalize(float const*, float const*): error: undefined reference to '__kmpc_fork_call' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined.: error: undefined reference to '__kmpc_for_static_init_4' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined.: error: undefined reference to '__kmpc_for_static_fini' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined..1: error: undefined reference to '__kmpc_for_static_init_4' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined..1: error: undefined reference to '__kmpc_for_static_fini' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined..2: error: undefined reference to '__kmpc_for_static_init_4' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined..2: error: undefined reference to '__kmpc_for_static_fini' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function ncnn::copy_make_border(ncnn::Mat const&, ncnn::Mat&, int, int, int, int, int, float): error: undefined reference to '__kmpc_fork_call' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined..3: error: undefined reference to '__kmpc_for_static_init_4' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function .omp_outlined..3: error: undefined reference to '__kmpc_for_static_fini' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function ncnn::copy_cut_border(ncnn::Mat const&, ncnn::Mat&, int, int, int, int): error: undefined reference to '__kmpc_fork_call' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(mat.cpp.o):/home/nihui/osd/ncnn/src/mat.cpp:function ncnn::resize_bilinear(ncnn::Mat const&, ncnn::Mat&, int, int): error: undefined reference to '__kmpc_fork_call' ../../../../src/main/jniLibs/armeabi-v7a/libncnn.a(convolutiondepthwise.cpp.o):/home/nihui/osd/ncnn/src/layer/convolutiondepthwise.cpp:function .omp_outlined..1: error: undefined reference to '__kmpc_for_static_init_8' collect2: error: ld returned 1 exit status ninja: build stopped: subcommand failed.

新版本编译出现这个错误， openmp的问题，你遇到没

[abc1225]

gradle里配置的编译器把gcc修改为clang，arguments "-DANDROID_TOOLCHAIN=clang"

[rivercn]

@abc1225 搞定了，非常感谢！ 测试4线程时整体速度是提升了，但是单线程模式的速度感觉反而降低的，准备换手机测试看!

[abc1225]

@rivercn 测完了告诉一下哈

[moli232777144]

之前尝试过更新版本的，新版需设置clang编译了，我这边尝试过，一直报下面这个错误，没找到解决方法，你们还修改过其他编译细节吗？或者是我需要更新软件某个版本？麻烦告知下； Error:cannot find -lomp Error:error: linker command failed with exit code 1 (use -v to see invocation)

[abc1225]

@moli232777144 没其它细节了，仅替换了lib库和头文件，gcc修改为clang，你这个的只能是ndk版本导致的问题，ndk版本号： 15.1.4xxx

[rivercn]

@abc1225 初步测试单线程是有上升，单张图片的时间跟视频的时间有差别，需要进一步验证

[abc1225]

@rivercn thanks

[moli232777144]

更新了相关可调参数的接口和测试数据，个人机子上测试，速度反而没之前版本快；mtcnn通过框架提升速度的空间可能不大了，模型计算量并不大，波动反而可能更大；欢迎测试交流下；

[Roger8]

@abc1225 ，已按照你说的修改，请问这个是什么原因 Information:Gradle tasks [:app:assembleDebug] Error:error: linker command failed with exit code 1 (use -v to see invocation)

[moli232777144]

@Roger8 升级ndk

[Roger8]

ndk版本由17换为r15后编译成功,3q

[arvrschool]

有多快呢？我在Android上跑人脸检测，很不稳定，每帧耗时波动比较大

[pxEkin]

更新了相关可调参数的接口和测试数据，个人机子上测试，速度反而没之前版本快；mtcnn通过框架提升速度的空间可能不大了，模型计算量并不大，波动反而可能更大；欢迎测试交流下；

嗯，我这边测试也是，用ncnn官方编译库，20180427的库比20180830的库还要快一些

[pxEkin]

有多快呢？我在Android上跑人脸检测，很不稳定，每帧耗时波动比较大

我跑也有几十毫秒到上百毫秒的波动

[galileolvpo]

我测下来线程越多，速度越慢....

[Captain1986]

之前尝试过更新版本的，新版需设置clang编译了，我这边尝试过，一直报下面这个错误，没找到解决方法，你们还修改过其他编译细节吗？或者是我需要更新软件某个版本？麻烦告知下； Error:cannot find -lomp Error:error: linker command failed with exit code 1 (use -v to see invocation)

openmp？