New Version Compile Issues!

[liuzhi136]

build/src/operator/rnn.o: In function mxnet::op::RNNProp::CreateOperatorEx(mxnet::Context, std::vector<nnvm::TShape, std::allocator<nnvm::TShape> >*, std::vector<int, std::allocator<int> >*) const': rnn.cc:(.text+0x4ee71): undefined reference to mxnet::Operator* mxnet::op::CreateOpmshadow::gpu(mxnet::op::RNNParam, int)'

I try to update mxnet on my computer to the current newest version. But I got a lot of errors like above.
I use cuda 8 and cudnn 5.1 during compiling. I clone the source code from the mxnet github website. Does any one meet the same issue like me ? and help out!

[vdantu]

@liuzhi136 : Just curious, would #7072 be solving the issue?
@mxnet-label-bot : [C++, Question]

[liuzhi136]

@vdantu The issue you offered was help, which apart of mxnet's file can be compiled. But it still encountered the same problem for the rest part of mxnet. It looks like these:
a - build/src/operator/convolution_v1_gpu.o
a - build/src/operator/correlation_gpu.o
a - build/src/operator/crop_gpu.o
a - build/src/operator/grid_generator_gpu.o
a - build/src/operator/identity_attach_KL_sparse_reg_gpu.o
a - build/src/operator/instance_norm_gpu.o
a - build/src/operator/l2_normalization_gpu.o
a - build/src/operator/leaky_relu_gpu.o
a - build/src/operator/loss_binary_op_gpu.o
a - build/src/operator/make_loss_gpu.o
a - build/src/operator/optimizer_op_gpu.o
a - build/src/operator/pad_gpu.o
a - build/src/operator/pooling_v1_gpu.o
a - build/src/operator/regression_output_gpu.o
a - build/src/operator/rnn_gpu.o
a - build/src/operator/roi_pooling_gpu.o
a - build/src/operator/sequence_last_gpu.o
a - build/src/operator/sequence_mask_gpu.o
a - build/src/operator/sequence_reverse_gpu.o
a - build/src/operator/slice_channel_gpu.o
a - build/src/operator/softmax_output_gpu.o
a - build/src/operator/spatial_transformer_gpu.o
a - build/src/operator/svm_output_gpu.o
a - build/src/operator/swapaxis_gpu.o

The rest still have the same problem.
build/src/operator/contrib/ctc_loss_gpu.o: In function mxnet::op::CTCLossOp<mshadow::gpu>::~CTCLossOp()': tmpxft_00000e38_00000000-4_ctc_loss.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEED2Ev[_ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEED5Ev]+0x50): undefined reference to cudnnDestroyCTCLossDescriptor'
build/src/operator/contrib/ctc_loss_gpu.o: In function mxnet::op::CTCLossOp<mshadow::gpu>::CTCLossOp(mxnet::op::CTCLossParam)': tmpxft_00000e38_00000000-4_ctc_loss.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEEC2ENS0_12CTCLossParamE[_ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEEC5ENS0_12CTCLossParamE]+0x4a): undefined reference to cudnnCreateCTCLossDescriptor'
tmpxft_00000e38_00000000-4_ctc_loss.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEEC2ENS0_12CTCLossParamE[_ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEEC5ENS0_12CTCLossParamE]+0x122): undefined reference to cudnnSetCTCLossDescriptor' build/src/operator/contrib/ctc_loss_gpu.o: In function mxnet::op::CTCLossOpmshadow::gpu::cudnn_forward(mxnet::OpContext const&, mshadow::Streammshadow::gpu, mshadow::Tensor<mshadow::gpu, 3, float>, mshadow::Tensor<mshadow::gpu, 1, float>, mshadow::Tensor<mshadow::gpu, 3, float>, std::vector<int, std::allocator >, std::vector<int, std::allocator >, std::vector<int, std::allocator >, int, int, int, bool)':
tmpxft_00000e38_00000000-4_ctc_loss.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEE13cudnn_forwardERKNS_9OpContextEPNS2_6StreamIS3_EENS2_6TensorIS3_Li3EfEENSB_IS3_Li1EfEESC_PSt6vectorIiSaIiEESH_SH_iiib[_ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEE13cudnn_forwardERKNS_9OpContextEPNS2_6StreamIS3_EENS2_6TensorIS3_Li3EfEENSB_IS3_Li1EfEESC_PSt6vectorIiSaIiEESH_SH_iiib]+0x30a): undefined reference to cudnnGetCTCLossWorkspaceSize' tmpxft_00000e38_00000000-4_ctc_loss.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEE13cudnn_forwardERKNS_9OpContextEPNS2_6StreamIS3_EENS2_6TensorIS3_Li3EfEENSB_IS3_Li1EfEESC_PSt6vectorIiSaIiEESH_SH_iiib[_ZN5mxnet2op9CTCLossOpIN7mshadow3gpuEE13cudnn_forwardERKNS_9OpContextEPNS2_6StreamIS3_EENS2_6TensorIS3_Li3EfEENSB_IS3_Li1EfEESC_PSt6vectorIiSaIiEESH_SH_iiib]+0x752): undefined reference to cudnnCTCLoss'
build/src/operator/nn/convolution_gpu.o: In function MaxForwardAlgos(cudnnContext*)': tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._Z15MaxForwardAlgosP12cudnnContext[_Z15MaxForwardAlgosP12cudnnContext]+0x39): undefined reference to cudnnGetConvolutionForwardAlgorithmMaxCount'
build/src/operator/nn/convolution_gpu.o: In function MaxBackwardFilterAlgos(cudnnContext*)': tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._Z22MaxBackwardFilterAlgosP12cudnnContext[_Z22MaxBackwardFilterAlgosP12cudnnContext]+0x39): undefined reference to cudnnGetConvolutionBackwardFilterAlgorithmMaxCount'
build/src/operator/nn/convolution_gpu.o: In function MaxBackwardDataAlgos(cudnnContext*)': tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._Z20MaxBackwardDataAlgosP12cudnnContext[_Z20MaxBackwardDataAlgosP12cudnnContext]+0x39): undefined reference to cudnnGetConvolutionBackwardDataAlgorithmMaxCount'
build/src/operator/nn/convolution_gpu.o: In function mxnet::op::CuDNNConvolutionOp<float>::InitDescriptors(std::vector<nnvm::TShape, std::allocator<nnvm::TShape> > const&, std::vector<nnvm::TShape, std::allocator<nnvm::TShape> > const&, cudnnDataType_t, cudnnDataType_t)': tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_[_ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_]+0x28d): undefined reference to cudnnSetConvolutionMathType'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA[ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA]+0x35d): undefined reference to cudnnSetConvolutionMathType' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_[_ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_]+0x42d): undefined reference to cudnnSetConvolutionMathType'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA[ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA]+0x4fc): undefined reference to cudnnSetConvolutionGroupCount' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_[_ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_]+0x5cb): undefined reference to cudnnSetConvolutionGroupCount'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA[ZN5mxnet2op18CuDNNConvolutionOpIfE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA]+0x69a): undefined reference to cudnnSetConvolutionGroupCount' build/src/operator/nn/convolution_gpu.o: In function mxnet::op::CuDNNConvolutionOp::InitDescriptors(std::vector<nnvm::TShape, std::allocatornnvm::TShape > const&, std::vector<nnvm::TShape, std::allocatornnvm::TShape > const&, cudnnDataType_t, cudnnDataType_t)':
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA[ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA]+0x28d): undefined reference to cudnnSetConvolutionMathType' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_[_ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_]+0x35d): undefined reference to cudnnSetConvolutionMathType'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA[ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA]+0x42d): undefined reference to cudnnSetConvolutionMathType' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_[_ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_]+0x4fc): undefined reference to cudnnSetConvolutionGroupCount'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA[ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA]+0x5cb): undefined reference to cudnnSetConvolutionGroupCount' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_[_ZN5mxnet2op18CuDNNConvolutionOpIdE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS5_EES9_15cudnnDataType_tSA_]+0x69a): undefined reference to cudnnSetConvolutionGroupCount'
build/src/operator/nn/convolution_gpu.o: In function mxnet::op::CuDNNConvolutionOp<mshadow::half::half_t>::InitDescriptors(std::vector<nnvm::TShape, std::allocator<nnvm::TShape> > const&, std::vector<nnvm::TShape, std::allocator<nnvm::TShape> > const&, cudnnDataType_t, cudnnDataType_t)': tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_[_ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_]+0x28d): undefined reference to cudnnSetConvolutionMathType'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD[ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD]+0x35d): undefined reference to cudnnSetConvolutionMathType' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_[_ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_]+0x42d): undefined reference to cudnnSetConvolutionMathType'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD[ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD]+0x4fc): undefined reference to cudnnSetConvolutionGroupCount' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_[_ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_]+0x5cb): undefined reference to cudnnSetConvolutionGroupCount'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD[ZN5mxnet2op18CuDNNConvolutionOpIN7mshadow4half6half_tEE15InitDescriptorsERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD]+0x69a): undefined reference to cudnnSetConvolutionGroupCount' build/src/operator/nn/convolution_gpu.o: In function mxnet::op::CuDNNConvolutionOp::SelectAlgo(mxnet::RunContext const&, std::vector<nnvm::TShape, std::allocatornnvm::TShape > const&, std::vector<nnvm::TShape, std::allocatornnvm::TShape > const&, cudnnDataType_t, cudnnDataType_t)':
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD[ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD]+0x18d): undefined reference to cudnnSetConvolutionMathType' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_[_ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_]+0x270): undefined reference to cudnnSetConvolutionMathType'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD[ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD]+0x34f): undefined reference to cudnnSetConvolutionMathType' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_[_ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_]+0x69a): undefined reference to cudnnGetConvolutionForwardAlgorithm_v7'
tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text.ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD[ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD]+0xc6b): undefined reference to cudnnGetConvolutionBackwardFilterAlgorithm_v7' tmpxft_0000210a_00000000-4_convolution.compute_61.cudafe1.cpp:(.text._ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_[_ZN5mxnet2op18CuDNNConvolutionOpIfE10SelectAlgoERKNS_10RunContextERKSt6vectorIN4nnvm6TShapeESaIS8_EESC_15cudnnDataType_tSD_]+0xf1c): undefined reference to cudnnGetConvolutionBackwardDataAlgorithm_v7'
build/src/operator/nn/convolution_gpu.o: In function `mxnet::op::CuDNNConvolutionOp::SelectAlgo(mxnet::RunContext const&, std::vector<nnvm::TShape, std::allocatornnvm::TShape > const&, std::vector<nnvm::TShape, std::allocatornnvm::TShape > const&, cudnnDataType_t, cudnnDataType_t)':

[leleamol]

Hi @liuzhi136 ,
If you are building on Ubuntu, the build from source instructions on the links below would be helpful.

1. http://mxnet.incubator.apache.org/install/ubuntu_setup.html
2. http://mxnet.incubator.apache.org/install/build_from_source.html

The installation links mention Cuda 9.2 and CuDNN 7.14 versions.
I have tried these instructions on my setup and could build the mxnet without errors.

Please provide the build flags or config.mk file , it will help in root causing the issue.

[leleamol]

Hi @liuzhi136 ,

Please let us know if you were able to build the cpp-package after following documents mentioned in the last comment.
@mxnet-label-bot [pending-requestor-info]

[leleamol]

@mxnet-label-bot add [pending-requester-info]

[liuzhi136]

@leleamol No. It does not help. I tried these instructions. I don't no why this building process failed. I had successfully build mxnet from source before. Maybe it's some inner errors stemming from my ubuntu system.

[leleamol]

The issue doesn't seem to be related to C++ API. WIll remove the C++ label.

@liuzhi136 can you please upload the errors that you are getting when building from source. Also please share the config flags that you have used to build the mxnet.

[leleamol]

@mxnet-label-bot update [Pending Requester Info, Question, Build, Ubuntu]

[vdantu]

@liuzhi136 : Could you upload the config flags you are using? Are you still facing this issue? If you can attach the config.mk file here, we can try and recreate this issue.