Package Details: opencv-cuda 4.4.0-2

Git Clone URL: https://aur.archlinux.org/opencv-cuda.git (read-only, click to copy)
Package Base: opencv-cuda
Description: Open Source Computer Vision Library with CUDA support
Upstream URL: http://opencv.org/
Licenses: BSD
Conflicts: opencv, opencv-samples
Provides: opencv, opencv-samples
Submitter: billypilgrim
Maintainer: billypilgrim
Last Packager: billypilgrim
Votes: 9
Popularity: 0.167477
First Submitted: 2018-08-11 08:14
Last Updated: 2020-08-02 12:59

Dependencies (24)

Required by (292)

Sources (2)

Latest Comments

« First ‹ Previous 1 2 3 4 5 Next › Last »

znmeb commented on 2020-03-02 23:47

CMake Error at modules/dnn/CMakeLists.txt:99 (message):
  CUDA backend for DNN module requires CC 5.3 or higher.  Please remove
  unsupported architectures from CUDA_ARCH_BIN option.


-- Configuring incomplete, errors occurred!
See also "/home/znmeb/.cache/yay/opencv-cuda/src/build/CMakeFiles/CMakeOutput.log".
See also "/home/znmeb/.cache/yay/opencv-cuda/src/build/CMakeFiles/CMakeError.log".
==> ERROR: A failure occurred in build().
    Aborting...
Error making: opencv-cuda

billypilgrim commented on 2020-03-01 10:04

@petronny I've pushed a fix, so hopefully it should work for you now.

petronny commented on 2020-03-01 05:06

==> Starting prepare()...
  -> Patching sources for CUDA v10
patching file modules/core/include/opencv2/core/cvstd_wrapper.hpp
patching file modules/cudacodec/CMakeLists.txt
patching file modules/cudacodec/src/cuvid_ffmpeg_api.hpp
patching file modules/cudacodec/src/cuvid_ffmpeg_impl.hpp
patching file modules/cudacodec/src/precomp.hpp
Hunk #1 FAILED at 81.
1 out of 1 hunk FAILED -- saving rejects to file modules/cudacodec/src/precomp.hpp.rej
==> ERROR: A failure occurred in prepare().
    Aborting...

A patch failed in a clean build. Please fix it.

sl1pkn07 commented on 2020-02-14 00:01

fix-cuda-10.1.patch seems not need anymore. build OK without it

greetings

sl1pkn07 commented on 2020-02-13 19:00

this part

/usr/bin/ld: /usr/lib/libIlmImf-2_4.so.24: undefined reference to `std::__cxx11::basic_stringstream<char, std::char_traits<char>, std::allocator<char> >::basic_stringstream()@GLIBCXX_3.4.26'
/usr/bin/ld: /usr/lib/libopencv_imgcodecs.so.4.2.0: undefined reference to `std::basic_stringstream<char, std::char_traits<char>, std::allocator<char> >::basic_stringstream()@GLIBCXX_3.4.26'
/usr/bin/ld: 3rdparty/dmlc-core/libdmlc.so: undefined reference to `std::__cxx11::basic_ostringstream<char, std::char_traits<char>, std::allocator<char> >::basic_ostringstream()@GLIBCXX_3.4.26'

is gone if build openexr and opencv-cuda with GCC 8 with

    -DCMAKE_C_COMPILER=/usr/bin/cc-8 \
    -DCMAKE_C_COMPILER_AR=/usr/bin/gcc-ar-8 \
    -DCMAKE_C_COMPILER_RANLIB=/usr/bin/gcc-ranlib-8 \
    -DCMAKE_CXX_COMPILER=/usr/bin/c++-8 \
    -DCMAKE_CXX_COMPILER_AR=/usr/bin/gcc-ar-8 \
    -DCMAKE_CXX_COMPILER_RANLIB=/usr/bin/gcc-ranlib-8

the rest still remain

EDIT: fixed. in one of my test i use -D_GLIBCXX_USE_CXX11_ABI, remove it, but keep the gcc-8 build, all OK

greetings

billypilgrim commented on 2020-02-13 17:49

@sl1pkn07 Hmm yeah I think it might be something to do with mxnet, because I'm not getting linking errors myself.

I have had linking errors in the past that happen when the list of libs to link against is in the wrong order so that could be what's happening here. It looks like it's trying and failing to link against libstdc++.so, which can also happen with clang if you don't link against it explicitly.

sl1pkn07 commented on 2020-02-13 16:38

build ok. but when try to build mxnet fails in linker step, but i'm not sure if is a opencv related

https://sl1pkn07.wtf/paste/view/8ff92321

greetings

billypilgrim commented on 2020-02-12 22:04

gcc-8 is being used as the CUDA host compiler already so the CUDA bits should be fine.

@sl1pkn07 What issue are you having? Will opencv-cuda not build for you or does it not work properly?

sl1pkn07 commented on 2020-02-12 20:51

@petronny what happen in mxnet? i have same issue, but i'm not sure if is the problem if use gcc-8 or not

billypilgrim commented on 2020-02-06 10:02

I'm not sure I understand. Yes, CUDA needs gcc8, but there's no harm in building this package with gcc9. Which compiler OpenCV is built won't change the ABI. If it did, every time there was a new version of gcc then your system would break. What's the issue you're having with mxnet?