Skip to content

fix multi-gpu build#616

Open
dskhudia wants to merge 1 commit intoNVIDIA:mainfrom
dskhudia:multi_gpu_build_fix
Open

fix multi-gpu build#616
dskhudia wants to merge 1 commit intoNVIDIA:mainfrom
dskhudia:multi_gpu_build_fix

Conversation

@dskhudia
Copy link
Contributor

Multi-gpu build with the following command fails on Ubuntu 20.04.5. Fix it by linking mpi_cxx library.

cmake -DSM=80 -DCMAKE_BUILD_TYPE=Release -DBUILD_PYT=ON -DBUILD_MULTI_GPU=ON ..; make -j

/usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o: in function MPI::Op::Init(void ()(void const, void*, int, MPI::Datatype const&), bool)':
bert_triton_example.cc:(.text._ZN3MPI2Op4InitEPFvPKvPviRKNS_8DatatypeEEb[_ZN3MPI2Op4InitEPFvPKvPviRKNS_8DatatypeEEb]+0x1d): undefined reference to ompi_mpi_cxx_op_intercept' /usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o: in function MPI::Intracomm::Clone() const':
bert_triton_example.cc:(.text._ZNK3MPI9Intracomm5CloneEv[_ZNK3MPI9Intracomm5CloneEv]+0x40): undefined reference to MPI::Comm::Comm()' /usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o: in function MPI::Graphcomm::Clone() const':
bert_triton_example.cc:(.text._ZNK3MPI9Graphcomm5CloneEv[_ZNK3MPI9Graphcomm5CloneEv]+0x3b): undefined reference to MPI::Comm::Comm()' /usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o: in function MPI::Cartcomm::Sub(bool const*) const':
bert_triton_example.cc:(.text._ZNK3MPI8Cartcomm3SubEPKb[_ZNK3MPI8Cartcomm3SubEPKb]+0xa1): undefined reference to MPI::Comm::Comm()' /usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o: in function MPI::Intracomm::Create_graph(int, int const*, int const*, bool) const':
bert_triton_example.cc:(.text._ZNK3MPI9Intracomm12Create_graphEiPKiS2_b[_ZNK3MPI9Intracomm12Create_graphEiPKiS2_b]+0x42): undefined reference to MPI::Comm::Comm()' /usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o: in function MPI::Cartcomm::Clone() const':
bert_triton_example.cc:(.text._ZNK3MPI8Cartcomm5CloneEv[_ZNK3MPI8Cartcomm5CloneEv]+0x3b): undefined reference to MPI::Comm::Comm()' /usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o:bert_triton_example.cc:(.text._ZNK3MPI9Intracomm11Create_cartEiPKiPKbb[_ZNK3MPI9Intracomm11Create_cartEiPKiPKbb]+0xaa): more undefined references to MPI::Comm::Comm()' follow
/usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o:(.data.rel.ro._ZTVN3MPI8DatatypeE[_ZTVN3MPI8DatatypeE]+0x78): undefined reference to MPI::Datatype::Free()' /usr/bin/ld: CMakeFiles/bert_triton_example.dir/bert_triton_example.cc.o:(.data.rel.ro._ZTVN3MPI3WinE[_ZTVN3MPI3WinE]+0x48): undefined reference to MPI::Win::Free()'`

@dskhudia
Copy link
Contributor Author

@byshiue for review. Is this an issue you have seen?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant