TMVA Compilation error in ROOT 6.16


_ROOT Version: 6.16
_Platform: Linux 2.6.32-504.8.1.el6.x86_64 #1 SMP Tue Jan 27 13:39:10 CST 2015 x86_64 x86_64 x86_64 GNU/Linux
_Compiler: g++ (GCC) 5.1.0


I get the following compilation error when trying to compile ROOT 6.16

[ 94%] Building CXX object tmva/tmva/CMakeFiles/TMVA.dir/src/DNN/Architectures/Cpu/CpuMatrix.cxx.o
[ 95%] Built target HistFactory
[ 96%] Built target Eve
[ 96%] Built target hist2workspace
In file included from /home/zar30002/gitlab/root/build/include/TMVA/DNN/Architectures/Cpu.h:21:0,
                 from /home/zar30002/gitlab/root/tmva/tmva/src/DNN/Architectures/Cpu.cxx:16:
/home/zar30002/gitlab/root/build/include/TMVA/DNN/Functions.h: In substitution of ‘template<class Architecture_t> decltype (Architecture_t:: CrossEntropy(Y, output, weights)) TMVA::DNN::evaluate(TMVA::DNN::ELossFunction, const typename Architecture_t::Matrix_t&, const typename Architecture_t::Matrix_t&, const typename Architecture_t::Matrix_t&) [with Architecture_t = TMVA::DNN::TCpu<double>]’:
/home/zar30002/gitlab/root/tmva/tmva/src/DNN/Architectures/Cpu/Propagation.cxx:332:30:   required from ‘TMVA::DNN::TCpu<AReal>::ConvLayerForward(std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, const std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, const TMVA::DNN::TCpuMatrix<AReal>&, const TMVA::DNN::TCpuMatrix<AReal>&, const TConvParams&, TMVA::DNN::EActivationFunction, std::vector<TMVA::DNN::TCpuMatrix<AReal> >&)::<lambda(UInt_t)> [with AFloat = double; UInt_t = unsigned int’
/home/zar30002/gitlab/root/tmva/tmva/src/DNN/Architectures/Cpu/Propagation.cxx:331:57:   required from ‘struct TMVA::DNN::TCpu<AReal>::ConvLayerForward(std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, const std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, const TMVA::DNN::TCpuMatrix<AReal>&, const TMVA::DNN::TCpuMatrix<AReal>&, const TConvParams&, TMVA::DNN::EActivationFunction, std::vector<TMVA::DNN::TCpuMatrix<AReal> >&) [with AReal = double; TMVA::DNN::CNN::TConvParams = TMVA::DNN::CNN::TConvParams]::<lambda(UInt_t)>’
/home/zar30002/gitlab/root/tmva/tmva/src/DNN/Architectures/Cpu/Propagation.cxx:334:4:   required from ‘static void TMVA::DNN::TCpu<AReal>::ConvLayerForward(std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, const std::vector<TMVA::DNN::TCpuMatrix<AReal> >&, const TMVA::DNN::TCpuMatrix<AReal>&, const TMVA::DNN::TCpuMatrix<AReal>&, const TConvParams&, TMVA::DNN::EActivationFunction, std::vector<TMVA::DNN::TCpuMatrix<AReal> >&) [with AReal = double; TMVA::DNN::CNN::TConvParams = TMVA::DNN::CNN::TConvParams]’
/home/zar30002/gitlab/root/tmva/tmva/src/DNN/Architectures/Cpu.cxx:30:19:   required from here
/home/zar30002/gitlab/root/build/include/TMVA/DNN/Functions.h:178:44: internal compiler error: Segmentation fault
    -> decltype(Architecture_t::CrossEntropy(Y, output, weights))
                                            ^
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
make[2]: *** [tmva/tmva/CMakeFiles/TMVA.dir/src/DNN/Architectures/Cpu.cxx.o] Error 1
make[1]: *** [tmva/tmva/CMakeFiles/TMVA.dir/all] Error 2

Hi Javier,
there seems to be a problem with your compiler. What platform are you on? I see your kernel is quite old.

Switching to a newer compiler will probably fix the issue – or you can download one of our precompiled binaries in case they match your platform/compiler.

Cheers,
Enrico

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.