I'm trying to build opencv' with cuda 10.2. When the following command :
cmake -DCMAKE_BUILD_TYPE=RELEASE \
-DOPENCV_EXTRA_MODULES_PATH=../../opencv_contrib/modules \
-DWITH_TBB=ON -DWITH_CUDA=ON \
-DBUILD_opencv_cudacodec=OFF \
-DENABLE_FAST_MATH=1 \
-DWITH_CUBLAS=1 \
-DWITH_V4L=ON \
-DWITH_OPENGL=ON \
-DWITH_GSTREAMER=ON \
-DOPENCV_GENERATE_PKGCONFIG=ON \
-DOPENCV_ENABLE_NONFREE=ON \
-DBUILD_EXAMPLES=TRUE \
-DBUILD_PERF_TESTS=FALSE \
-DEBUILD_TESTS=FALSE ../../opencv
I have the following issue :
Could NOT find CUDNN (missing: CUDNN_LIBRARY CUDNN_INCLUDE_DIR) (Required is at least version "6")
of course I have installed cudnn7 corresponding to cuda 10.2, I the installation test passed.
Can someone help ?
To fix the problem I added the following options for cmake :
-DCUDNN_INCLUDE_DIR=/usr/local/cuda/include \
-DCUDNN_LIBRARY=/usr/local/cuda/lib64/libcudnn.so.7.6.5 \
You may also need to add:
-DCUDNN_VERSION='7.6'
Or in my case it was:
-DCUDNN_VERSION='8.0'
Related
I am trying to compile the source code for Inkscape following this guide https://wiki.inkscape.org/wiki/index.php/Compiling_Inkscape_on_Windows_with_MSYS2
I get the following errors
-- Checking for modules 'harfbuzz;pangocairo;pangoft2;fontconfig;gsl;gmodule-2.0;libsoup-2.4>=2.42
-- No package 'harfbuzz' found
-- No package 'pangocairo' found
-- No package 'pangoft2' found
-- No package 'fontconfig' found
-- No package 'gsl' found
-- No package 'gmodule-2.0' found
-- No package 'libsoup-2.4' found
I found this answer Error while compiling Inkscape from source code but that only works for Ubuntu and I am working on Windows.
I ran this script
msys2installdeps.sh
# install Inkscape dependencies (required)
eval pacman -S $PACMAN_OPTIONS \
$ARCH-double-conversion \
$ARCH-gc \
$ARCH-gsl \
$ARCH-libxslt \
$ARCH-boost \
$ARCH-gtk3 \
$ARCH-gtkmm3 \
$ARCH-gdl \
$ARCH-libsoup
# install Inkscape dependencies (optional)
eval pacman -S $PACMAN_OPTIONS \
$ARCH-poppler \
$ARCH-potrace \
$ARCH-libcdr \
$ARCH-libvisio \
$ARCH-libwpg \
$ARCH-aspell \
$ARCH-aspell-en \
$ARCH-gtkspell3
# install ImageMagick (as Inkscape requires old version ImageMagick 6 we have to specify it explicitly)
# to prevent future updates:
# add the line
# "IgnorePkg = mingw-w64-*-imagemagick"
# to
# "C:\msys64\etc\pacman.conf"
# or (always!) run pacman with the additional command line switch
# --ignore=mingw-w64-*-imagemagick
for arch in $(eval echo $ARCH); do
wget -nv https://gitlab.com/ede123/bintray/raw/master/${arch}-imagemagick-6.9.10.69-1-any.pkg.tar.xz \
&& pacman -U $PACMAN_OPTIONS ${arch}-imagemagick-6.9.10.69-1-any.pkg.tar.xz \
&& rm ${arch}-imagemagick-6.9.10.69-1-any.pkg.tar.xz
done
# install Python and modules used by Inkscape
eval pacman -S $PACMAN_OPTIONS \
$ARCH-python3 \
$ARCH-python3-pip \
$ARCH-python3-lxml \
$ARCH-python3-numpy \
$ARCH-python3-pillow \
$ARCH-python3-six \
$ARCH-python3-gobject
PACKAGES="coverage pyserial scour"
for arch in $(eval echo $ARCH); do
case ${arch} in
mingw-w64-i686)
/mingw32/bin/pip3 install --upgrade --upgrade-strategy eager ${PACKAGES} || \
/mingw32/bin/pip3 install --upgrade ${PACKAGES}
;;
mingw-w64-x86_64)
C:/MinGW/opt/python3/Scripts/pip3 install --upgrade --upgrade-strategy eager ${PACKAGES} || \
C:/MinGW/opt/python3/Scripts/pip3 install --upgrade ${PACKAGES}
;;
esac
done
The script downloads and installs Inkscape's dependencies.
ndk build ffmpeg 3.4.2 configure success ,but when I make it .fatal error like this
configure script is
I make it in CentOS Linux release 7.4.1708 (Core)
similar, I make fdkaac fatal error too like this
I don't know I what is the problem of system
I think you should change android ndk version
I use the ndk r16b I get the problem also
so I use the ndk r15c
I success
I holp this can help you
this is my .sh file
#!/bin/bash
rm -rf android/arm
echo "stone"
NDK=/xxxxxxx/android-ndk-r15c
NDK_VERSION=android-14
ARCH=arm
SYSROOT=$NDK/platforms/$NDK_VERSION/arch-$ARCH
TOOLCHAIN=$NDK/toolchains/arm-linux-androideabi-4.9/prebuilt/darwin-x86_64
CROSS_COMPILE=$TOOLCHAIN/bin/arm-linux-androideabi-
CPU=arm
PREFIX=$(pwd)/android/$CPU
ADDI_CFLAGS="-marm"
function build_one
{
./configure \
--prefix=$PREFIX \
--enable-shared \
--disable-static \
--disable-doc \
--disable-ffmpeg \
--disable-ffplay \
--disable-ffprobe \
--disable-ffserver \
--disable-symver \
--cross-prefix=$CROSS_COMPILE \
--target-os=android \
--arch=$ARCH \
--enable-cross-compile \
--sysroot=$SYSROOT \
--extra-cflags="-Os -fpic $ADDI_CFLAGS" \
--extra-ldflags="$ADDI_LDFLAGS" \
$ADDITIONAL_CONFIGURE_FLAG
make clean
make -j8
make install
}
build_one
~~~~~~~~~~~~
I am trying to cross compile opencv with ffmpeg support for an arm board with a buildroot based custom linux. The host is an ubuntu PC.
I want static opencv libraries.
I have downloaded ffmpeg source and crosscompiled it with the following configuration
./configure \
--enable-cross-compile \
--cross-prefix=arm-linux-gnueabihf- \
--target-os=linux \
--arch=arm \
--disable-static \
--enable-shared \
--enable-nonfree \
--enable-ffmpeg \
--enable-gpl \
--enable-swscale \
--enable-pthreads \
--disable-yasm \
--disable-stripping \
--prefix=../build \
--extra-cflags=-I../build/include --extra-ldflags=-L../build/lib
The paths are all correct and ffmpeg is built successfully. Then when I try to configure opencv with cmake-gui, I have to manually specify all paths.
But despite this, opencv configuration is unable to resolve FFMPEG correctly
I have attached the screen shot below
It cannot resolve ffmpeg versions. I tried to ignore it and build but it fails with linker error
Linking CXX static library ../../lib/libopencv_features2d.a
[ 49%] Built target opencv_features2d
make: *** [all] Error 2
I'm getting the weirdest error when I do ./configure when building nmcurses-5.9
The issue is when I try to run
CC="gcc -m32" LD="ld -m32" ./configure \
--prefix=/mingw \
--without-cxx-binding \
--without-ada \
--enable-warnings \
--enable-assertions \
--enable-reentrant \
--with-debug \
--with-normal \
--disable-home-terminfo \
--enable-sp-funcs \
--enable-term-driver \
--enable-interop \
--with-pthread
and the error I get is
./configure: line 21016: D:\Program: No such file or directory
on that line is
${MAKE:-make} preinstall
I am building this in MinGW using msys.
any help would be apreciated.
In ${MAKE:-make} preinstall, the expression ${MAKE:-make} expands to the
value of the shell variable MAKE, if it is set, and otherwise to make.
So MAKE is set and expands to something probably of the form D:\Program Files\...,
i.e. a path with embedded spaces, which is construed as distinct tokens
D:\Program and Files\... by the shell when attempting to execute the
intended command:
\path\to\make preinstall
Instead it attempts to execute a program D:\Program with arguments Files\... preinstall
and complains that no such program exists.
With GNU autotools is advisable to install tools in paths that are free of
embedded spaces.
Mesa version = 12.0.1
Tried version before this as well.
At link time this error comes up: /usr/lib/gcc/x86_64-redhat-linux/4.8.5/../../../../lib64/libGL.so: undefined reference to _glapi_tls_Dispatch'.
The script i am using to install mesa:
#!/bin/bash
make -j4 distclean # if in an existing build
autoreconf -fi
./configure \
CXXFLAGS="-O2 -g -DDEFAULT_SOFTWARE_DEPTH_BITS=31" \
CFLAGS="-O2 -g -DDEFAULT_SOFTWARE_DEPTH_BITS=31" \
--disable-xvmc \
--disable-glx \
--disable-dri \
--with-dri-drivers="" \
--with-gallium-drivers="swrast" \
--enable-texture-float \
--enable-shared-glapi \
--disable-egl \
--with-egl-platforms="" \
--enable-gallium-osmesa \
--enable-llvm-shared-libs=yes \
--enable-gallium-llvm=yes \
--prefix=/home/ec2-user/new-mesa/mesaBuild/
make -j2
make -j4 install
After installation :
Added prefix(defined in script) to /etc/ld.so.conf' and ran ldconfig
Then in link command: -Lprefix -lOSMesa
What am i doing wrong?