Ask Your Question
0

Is there a CI server for OpenCV?

asked 2016-11-28 21:45:21 -0600

I'm trying to build the master branches of opencv (be0865406fc197809208229813c6f3e357e18857) and opencv_contrib (afd8a833c2188b8db525671520e5c64ebcdff232) on a Fedora 23 box with CUDA 8. I'm getting:

Scanning dependencies of target opencv_perf_cudaarithm
[ 49%] Building CXX object modules/cudaarithm/CMakeFiles/opencv_perf_cudaarithm.dir/perf/perf_reductions.cpp.o
[ 49%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_utils.cpp.o
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-Wno-delete-non-virtual-dtor’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-Wno-delete-non-virtual-dtor’ is valid for C++/ObjC++ but not for C
[ 49%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_rand.cpp.o
[ 50%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_mat.cpp.o
[ 50%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_intrin.cpp.o
[ 50%] Building CXX object modules/cudaarithm/CMakeFiles/opencv_perf_cudaarithm.dir/perf/perf_element_operations.cpp.o
[ 50%] Building NVCC (Device) object modules/cudafilters/CMakeFiles/cuda_compile.dir/src/cuda/cuda_compile_generated_row_filter.16sc4.cu.o
[ 50%] Building NVCC (Device) object modules/cudev/test/CMakeFiles/opencv_test_cudev.dir/opencv_test_cudev_generated_test_bitwize_op.cu.o
[ 50%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_countnonzero.cpp.o
[ 50%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_dxt.cpp.o
[ 50%] Building NVCC (Device) object modules/cudafilters/CMakeFiles/cuda_compile.dir/src/cuda/cuda_compile_generated_column_filter.16sc1.cu.o
[ 50%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_downhill_simplex.cpp.o
[ 50%] Building NVCC (Device) object modules/cudev/test/CMakeFiles/opencv_test_cudev.dir/opencv_test_cudev_generated_test_integral.cu.o
[ 50%] Building NVCC (Device) object modules/cudafilters/CMakeFiles/cuda_compile.dir/src/cuda/cuda_compile_generated_filter2d.cu.o
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-Wno-delete-non-virtual-dtor’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-std=c++11’ is valid for C++/ObjC++ but not for C
cc1: warning: command line option ‘-Wno-delete-non-virtual-dtor’ is valid for C++/ObjC++ but not for C
[ 50%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_misc.cpp.o
[ 50%] Building NVCC (Device) object modules/cudafilters/CMakeFiles/cuda_compile.dir/src/cuda/cuda_compile_generated_median_filter.cu.o
[ 50%] Building CXX object modules/core/CMakeFiles/opencv_test_core.dir/test/test_arithm.cpp.o
cc1: warning: command line option ‘-std=c ...
(more)
edit retag flag offensive close merge delete

Comments

yes, there is CI

unfortunately, it seems none of the boxes actually have CUDA

berak gravatar imageberak ( 2016-11-29 01:31:14 -0600 )edit

I can confirm that latest branches from both repos build fine with CUDA7.5, so it might be with the CUDA8 dependencies.

StevenPuttemans gravatar imageStevenPuttemans ( 2016-11-30 04:47:25 -0600 )edit
1

Thanks for the info about no CUDA capable machine in CI.

I tried CUDA 7.5 and it work, but there were updated to the master branch:

 3rdparty/openvx/include/ivx.hpp  modules/cudaarithm/perf/perf_arithm.cpp
 modules/cudaarithm/test/test_core.cpp
 modules/cudafeatures2d/test/test_features2d.cpp
 modules/cudafilters/test/test_filters.cpp
 modules/cudaimgproc/test/test_canny.cpp
 modules/cudaimgproc/test/test_histogram.cpp
 modules/cudaimgproc/test/test_hough.cpp
 modules/cudaobjdetect/test/test_objdetect.cpp 
...

When I tried the latest with CUDA 8 it worked!

pstieber gravatar imagepstieber ( 2016-11-30 12:54:18 -0600 )edit

1 answer

Sort by » oldest newest most voted
0

answered 2016-12-02 12:19:06 -0600

From the comments left above:

There are OpenCV CI servers, but they do not have GPUs so the CUDA builds are not generated on the CI servers.

edit flag offensive delete link more

Question Tools

1 follower

Stats

Asked: 2016-11-28 21:45:21 -0600

Seen: 646 times

Last updated: Dec 02 '16