<div dir="ltr"><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#000000">Seems <span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)">-isystem is a gcc option. </span></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#000000"><span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)">Can you try not to give "</span><span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)">-I/usr/</span><span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)">local/cuda-11.5/include" in CUDA_ARCH_FLAGS, see whether nvcc can find the proper include/ directory?</span></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#000000"><span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)"><br></span></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#000000"><span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)">Sherry</span></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Dec 21, 2021 at 10:26 AM Roland Richter <<a href="mailto:roland.richter@ntnu.no">roland.richter@ntnu.no</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
  
    
  
  <div>
    <p>I'm already using GCC, together with Intel MPI, thus that should
      not be the problem. But I'll take a look if there are ways to fix
      that problem, or to circumvent it without having to install
      superlu_dist separately.</p>
    <p>Thanks!</p>
    <p>Regards,</p>
    <p>Roland<br>
    </p>
    <div>Am 21.12.2021 um 19:04 schrieb Barry
      Smith:<br>
    </div>
    <blockquote type="cite">
      
      <br>
      <div>  I think the problem comes from the following: </div>
      <div><br>
      </div>
      <div>
        <blockquote type="cite">
          <div>
            <p>-DCUDA_ARCH_FLAGS="-I/usr/local/cuda-11.5/include
              -arch=sm_61 -DDEBUGlevel=0 -DPRNTlevel=0"</p>
          </div>
        </blockquote>
        which is ok, the location of the CUDA include files is passed in
        correctly but when the compile is done inside SuperLU_DIST's
        make</div>
      <div><br>
      </div>
      <div>cd
~/Downloads/git-files/petsc/mpich-complex-linux-gcc-demo/externalpackages/git.superlu_dist/petsc-build/SRC
        && /usr/local/cuda-11.5/bin/nvcc
        -forward-unknown-to-host-compiler -DSUPERLU_DIST_EXPORTS
        -Dsuperlu_dist_EXPORTS
-I~/Downloads/git-files/petsc/mpich-complex-linux-gcc-demo/externalpackages/git.superlu_dist/petsc-build/SRC
        -isystem=/usr/local/cuda-11.5/include -DUSE_VENDOR_BLAS
        -allow-unsupported-compiler -Xcompiler -fPIC -O3 -ccbin
        /opt/intel/oneapi/mpi/2021.5.0/bin/mpicxx -std=c++17 -gencode
        arch=compute_61,code=sm_61  -Wno-deprecated-gpu-targets
        -I/opt/intel/oneapi/mpi/2021.5.0/include -O3 -DNDEBUG
        --generate-code=arch=compute_61,code=[compute_61,sm_61]
        -Xcompiler=-fPIC -std=c++11 -MD -MT
        SRC/CMakeFiles/superlu_dist.dir/superlu_gpu_utils.cu.o -MF
        CMakeFiles/superlu_dist.dir/superlu_gpu_utils.cu.o.d -x cu -dc
~/Downloads/git-files/petsc/mpich-complex-linux-gcc-demo/externalpackages/git.superlu_dist/SRC/<a href="http://superlu_gpu_utils.cu" target="_blank">superlu_gpu_utils.cu</a>
        -o CMakeFiles/superlu_dist.dir/superlu_gpu_utils.cu.o</div>
      <div><br>
      </div>
      <div>The </div>
      <div><br>
      </div>
      <div> -I/usr/local/cuda-11.5/include</div>
      <div><br>
      </div>
      <div>has been replaced with </div>
      <div><br>
      </div>
      <div>-isystem=/usr/local/cuda-11.5/include </div>
      <div><br>
      </div>
      <div>and I suspect the nvcc doesn't understand it so it ignores
        it. I assume somewhere inside the CMAKE processing it is
        replacing the -I with the -isystem=</div>
      <div><br>
      </div>
      <div>I don't know how one could fix this.  I suggest you try
        installing SuperLU_DIST yourself directly. But I suspect the
        same problem will appear.</div>
      <div><br>
      </div>
      <div>It is possible that if you do not use the Intel compilers but
        use for example the GNU compilers the problem may not occur
        because cmake may not make the transformation.</div>
      <div><br>
      </div>
      <div>Barry</div>
      <div><br>
      </div>
      <div>Googling cmake isystem produces many messages about issues
        related to isystem and cmake but it would take a life-time to
        understand them.</div>
      <div><br>
      </div>
      <div><br>
        <blockquote type="cite">
          <div>On Dec 21, 2021, at 2:24 AM, Roland Richter <<a href="mailto:roland.richter@ntnu.no" target="_blank">roland.richter@ntnu.no</a>>
            wrote:</div>
          <br>
          <div>
            <div>
              <p>I added/replaced all six lines which are in
                that commit, getting <br>
              </p>
              <p>Executing: /usr/bin/cmake ..
                -DCMAKE_INSTALL_PREFIX=/opt/petsc
                -DCMAKE_INSTALL_NAME_DIR:STRING="/opt/petsc/lib"
                -DCMAKE_INSTALL_LIBDIR:STRING="lib"
                -DCMAKE_VERBOSE_MAKEFILE=1 -DCMAKE_BUILD_TYPE=Release
                -DCMAKE_C_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicc"
-DMPI_C_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicc"
                -DCMAKE_AR=/usr/bin/ar -DCMAKE_RANLIB=/usr/bin/ranlib
                -DCMAKE_C_FLAGS:STRING="-mavx2 -march=native -O3 -fPIC
                -fopenmp" -DCMAKE_C_FLAGS_DEBUG:STRING="-mavx2
                -march=native -O3 -fPIC -fopenmp"
                -DCMAKE_C_FLAGS_RELEASE:STRING="-mavx2 -march=native -O3
                -fPIC -fopenmp"
                -DCMAKE_CXX_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicxx"
-DMPI_CXX_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicxx"
                -DCMAKE_CXX_FLAGS:STRING="-mavx2 -march=native -O3
                -fopenmp -fPIC -std=gnu++17 -fopenmp"
                -DCMAKE_CXX_FLAGS_DEBUG:STRING="-mavx2 -march=native -O3
                -fopenmp -fPIC -std=gnu++17 -fopenmp"
                -DCMAKE_CXX_FLAGS_RELEASE:STRING="-mavx2 -march=native
                -O3 -fopenmp -fPIC -std=gnu++17 -fopenmp"
                -DCMAKE_Fortran_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpifc"
-DMPI_Fortran_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpifc"
                -DCMAKE_Fortran_FLAGS:STRING="-mavx2 -march=native -O3
                -fPIC -fopenmp -fallow-argument-mismatch"
                -DCMAKE_Fortran_FLAGS_DEBUG:STRING="-mavx2 -march=native
                -O3 -fPIC -fopenmp -fallow-argument-mismatch"
                -DCMAKE_Fortran_FLAGS_RELEASE:STRING="-mavx2
                -march=native -O3 -fPIC -fopenmp
                -fallow-argument-mismatch"
                -DCMAKE_EXE_LINKER_FLAGS:STRING=" -fopenmp -fopenmp"
                -DBUILD_SHARED_LIBS:BOOL=ON -DTPL_ENABLE_CUDALIB=TRUE
                -DTPL_CUDA_LIBRARIES="-Wl,-rpath,/usr/local/cuda-11.5/lib64
                -L/usr/local/cuda-11.5/lib64 -lcudart -lcufft -lcublas
                -lcusparse -lcusolver -lcurand
                -L/usr/local/cuda-11.5/lib64/stubs -lcuda"
                -DCUDA_ARCH_FLAGS="-I/usr/local/cuda-11.5/include
                -arch=sm_61 -DDEBUGlevel=0 -DPRNTlevel=0"
                -DCMAKE_CUDA_FLAGS="-allow-unsupported-compiler
                -Xcompiler -fPIC -O3 -ccbin
                /opt/intel/oneapi/mpi/2021.5.0/bin/mpicxx -std=c++17
                -gencode arch=compute_61,code=sm_61 
                -Wno-deprecated-gpu-targets
                -I"/opt/intel/oneapi/mpi/2021.5.0/include" "
                -DUSE_XSDK_DEFAULTS=YES -DTPL_BLAS_LIBRARIES="-lopenblas
                -Wl,-rpath,/usr/local/cuda-11.5/lib64
                -L/usr/local/cuda-11.5/lib64 -lcudart -lcufft -lcublas
                -lcusparse -lcusolver -lcurand
                -L/usr/local/cuda-11.5/lib64/stubs -lcuda -lm -lstdc++
                -ldl
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                -L/opt/intel/oneapi/mpi/2021.5.0/lib/release
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib
                -L/opt/intel/oneapi/mpi/2021.5.0/lib -lmpifort -lmpi
                -lrt -lpthread -lgfortran -lm
                -Wl,-rpath,/usr/lib64/gcc/x86_64-suse-linux/11
                -L/usr/lib64/gcc/x86_64-suse-linux/11
                -Wl,-rpath,/opt/intel/oneapi/vpl/2022.0.0/lib
                -L/opt/intel/oneapi/vpl/2022.0.0/lib
                -Wl,-rpath,/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
                -L/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                -L/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                -Wl,-rpath,/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                -L/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                -L/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                -L/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
-L/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
                -Wl,-rpath,/opt/intel/oneapi/dal/2021.5.1/lib/intel64
                -L/opt/intel/oneapi/dal/2021.5.1/lib/intel64
-Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
-L/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
                -Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                -L/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                -Wl,-rpath,/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                -L/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                -L/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                -Wl,-rpath,/usr/x86_64-suse-linux/lib
                -L/usr/x86_64-suse-linux/lib
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib -lgfortran
                -lm -lgcc_s -lquadmath"
                -DTPL_LAPACK_LIBRARIES="-lopenblas
                -Wl,-rpath,/usr/local/cuda-11.5/lib64
                -L/usr/local/cuda-11.5/lib64 -lcudart -lcufft -lcublas
                -lcusparse -lcusolver -lcurand
                -L/usr/local/cuda-11.5/lib64/stubs -lcuda -lm -lstdc++
                -ldl
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                -L/opt/intel/oneapi/mpi/2021.5.0/lib/release
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib
                -L/opt/intel/oneapi/mpi/2021.5.0/lib -lmpifort -lmpi
                -lrt -lpthread -lgfortran -lm
                -Wl,-rpath,/usr/lib64/gcc/x86_64-suse-linux/11
                -L/usr/lib64/gcc/x86_64-suse-linux/11
                -Wl,-rpath,/opt/intel/oneapi/vpl/2022.0.0/lib
                -L/opt/intel/oneapi/vpl/2022.0.0/lib
                -Wl,-rpath,/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
                -L/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                -L/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                -Wl,-rpath,/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                -L/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                -L/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                -L/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
-L/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
                -Wl,-rpath,/opt/intel/oneapi/dal/2021.5.1/lib/intel64
                -L/opt/intel/oneapi/dal/2021.5.1/lib/intel64
-Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
-L/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
                -Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                -L/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                -Wl,-rpath,/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                -L/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                -Wl,-rpath,/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                -L/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                -Wl,-rpath,/usr/x86_64-suse-linux/lib
                -L/usr/x86_64-suse-linux/lib
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib -lgfortran
                -lm -lgcc_s -lquadmath" -Denable_parmetislib=FALSE
                -DTPL_ENABLE_PARMETISLIB=FALSE -DXSDK_ENABLE_Fortran=ON
                -Denable_tests=0 -Denable_examples=0
                -DMPI_C_COMPILE_FLAGS:STRING=""
                -DMPI_C_INCLUDE_PATH:STRING=""
                -DMPI_C_HEADER_DIR:STRING="" -DMPI_C_LIBRARIES:STRING=""</p>
              <p>but the compilation still fails with the same
                error.<br>
              </p>
              <div>Am 20.12.21 um 22:46 schrieb
                Roland Richter:<br>
              </div>
              <blockquote type="cite">
                <p>Yes, just checked, I only included the
                  changes above the comment...</p>
                <p>Will test tomorrow, thanks for the help!</p>
                <p>Regards,</p>
                <p>Roland<br>
                </p>
                <div>Am 20.12.2021 um 21:46
                  schrieb Barry Smith:<br>
                </div>
                <blockquote type="cite">
                  <div>  Hmm, the fix should now be supplying a
                    -DCMAKE_CUDA_FLAGS to this cmake command (line 48
                    at <a href="https://gitlab.com/petsc/petsc/-/merge_requests/4635/diffs" target="_blank">https://gitlab.com/petsc/petsc/-/merge_requests/4635/diffs</a>)
                    I do not see that flag being set inside the
                    configure.log so I am guessing you didn't get the
                    complete fix.</div>
                  <div><br>
                  </div>
                  <div><br>
                  </div>
                  <div>Executing: /usr/bin/cmake ..
                    -DCMAKE_INSTALL_PREFIX=/opt/petsc
                    -DCMAKE_INSTALL_NAME_DIR:STRING="/opt/petsc/lib"
                    -DCMAKE_INSTALL_LIBDIR:STRING="lib"
                    -DCMAKE_VERBOSE_MAKEFILE=1
                    -DCMAKE_BUILD_TYPE=Release
                    -DCMAKE_C_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicc"
-DMPI_C_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicc"
                    -DCMAKE_AR=/usr/bin/ar
                    -DCMAKE_RANLIB=/usr/bin/ranlib
                    -DCMAKE_C_FLAGS:STRING="-mavx2 -march=native -O3
                    -fPIC -fopenmp" -DCMAKE_C_FLAGS_DEBUG:STRING="-mavx2
                    -march=native -O3 -fPIC -fopenmp"
                    -DCMAKE_C_FLAGS_RELEASE:STRING="-mavx2 -march=native
                    -O3 -fPIC -fopenmp"
                    -DCMAKE_CXX_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicxx"
-DMPI_CXX_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpicxx"
                    -DCMAKE_CXX_FLAGS:STRING="-mavx2 -march=native -O3
                    -fopenmp -fPIC -std=gnu++17 -fopenmp"
                    -DCMAKE_CXX_FLAGS_DEBUG:STRING="-mavx2 -march=native
                    -O3 -fopenmp -fPIC -std=gnu++17 -fopenmp"
                    -DCMAKE_CXX_FLAGS_RELEASE:STRING="-mavx2
                    -march=native -O3 -fopenmp -fPIC -std=gnu++17
                    -fopenmp"
                    -DCMAKE_Fortran_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpifc"
-DMPI_Fortran_COMPILER="/opt/intel/oneapi/mpi/2021.5.0/bin/mpifc"
                    -DCMAKE_Fortran_FLAGS:STRING="-mavx2 -march=native
                    -O3 -fPIC -fopenmp -fallow-argument-mismatch"
                    -DCMAKE_Fortran_FLAGS_DEBUG:STRING="-mavx2
                    -march=native -O3 -fPIC -fopenmp
                    -fallow-argument-mismatch"
                    -DCMAKE_Fortran_FLAGS_RELEASE:STRING="-mavx2
                    -march=native -O3 -fPIC -fopenmp
                    -fallow-argument-mismatch"
                    -DCMAKE_EXE_LINKER_FLAGS:STRING=" -fopenmp -fopenmp"
                    -DBUILD_SHARED_LIBS:BOOL=ON
                    -DTPL_ENABLE_CUDALIB=TRUE
                    -DTPL_CUDA_LIBRARIES="-Wl,-rpath,/usr/local/cuda-11.5/lib64
                    -L/usr/local/cuda-11.5/lib64 -lcudart -lcufft
                    -lcublas -lcusparse -lcusolver -lcurand
                    -L/usr/local/cuda-11.5/lib64/stubs -lcuda"
                    -DCUDA_ARCH_FLAGS="-I/usr/local/cuda-11.5/include
                    -arch=sm_61 -DDEBUGlevel=0 -DPRNTlevel=0"
                    -DUSE_XSDK_DEFAULTS=YES
                    -DTPL_BLAS_LIBRARIES="-lopenblas
                    -Wl,-rpath,/usr/local/cuda-11.5/lib64
                    -L/usr/local/cuda-11.5/lib64 -lcudart -lcufft
                    -lcublas -lcusparse -lcusolver -lcurand
                    -L/usr/local/cuda-11.5/lib64/stubs -lcuda -lm
                    -lstdc++ -ldl
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                    -L/opt/intel/oneapi/mpi/2021.5.0/lib/release
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib
                    -L/opt/intel/oneapi/mpi/2021.5.0/lib -lmpifort -lmpi
                    -lrt -lpthread -lgfortran -lm
                    -Wl,-rpath,/usr/lib64/gcc/x86_64-suse-linux/11
                    -L/usr/lib64/gcc/x86_64-suse-linux/11
                    -Wl,-rpath,/opt/intel/oneapi/vpl/2022.0.0/lib
                    -L/opt/intel/oneapi/vpl/2022.0.0/lib
                    -Wl,-rpath,/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
                    -L/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
-Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                    -L/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                    -Wl,-rpath,/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                    -L/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                    -L/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                    -L/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
-L/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
                    -Wl,-rpath,/opt/intel/oneapi/dal/2021.5.1/lib/intel64
                    -L/opt/intel/oneapi/dal/2021.5.1/lib/intel64
-Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
-L/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
                    -Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                    -L/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                    -Wl,-rpath,/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                    -L/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                    -L/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                    -Wl,-rpath,/usr/x86_64-suse-linux/lib
                    -L/usr/x86_64-suse-linux/lib
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib
                    -lgfortran -lm -lgcc_s -lquadmath"
                    -DTPL_LAPACK_LIBRARIES="-lopenblas
                    -Wl,-rpath,/usr/local/cuda-11.5/lib64
                    -L/usr/local/cuda-11.5/lib64 -lcudart -lcufft
                    -lcublas -lcusparse -lcusolver -lcurand
                    -L/usr/local/cuda-11.5/lib64/stubs -lcuda -lm
                    -lstdc++ -ldl
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                    -L/opt/intel/oneapi/mpi/2021.5.0/lib/release
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib
                    -L/opt/intel/oneapi/mpi/2021.5.0/lib -lmpifort -lmpi
                    -lrt -lpthread -lgfortran -lm
                    -Wl,-rpath,/usr/lib64/gcc/x86_64-suse-linux/11
                    -L/usr/lib64/gcc/x86_64-suse-linux/11
                    -Wl,-rpath,/opt/intel/oneapi/vpl/2022.0.0/lib
                    -L/opt/intel/oneapi/vpl/2022.0.0/lib
                    -Wl,-rpath,/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
                    -L/opt/intel/oneapi/tbb/2021.5.0/lib/intel64/gcc4.8
-Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                    -L/opt/intel/oneapi/mpi/2021.5.0/libfabric/lib
                    -Wl,-rpath,/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                    -L/opt/intel/oneapi/mkl/2022.0.1/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                    -L/opt/intel/oneapi/ipp/2021.5.1/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                    -L/opt/intel/oneapi/ippcp/2021.5.0/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
-L/opt/intel/oneapi/dnnl/2022.0.1/cpu_dpcpp_gpu_dpcpp/lib
                    -Wl,-rpath,/opt/intel/oneapi/dal/2021.5.1/lib/intel64
                    -L/opt/intel/oneapi/dal/2021.5.1/lib/intel64
-Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
-L/opt/intel/oneapi/compiler/2022.0.1/linux/compiler/lib/intel64_lin
                    -Wl,-rpath,/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                    -L/opt/intel/oneapi/compiler/2022.0.1/linux/lib
                    -Wl,-rpath,/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                    -L/opt/intel/oneapi/clck/2021.5.0/lib/intel64
                    -Wl,-rpath,/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                    -L/opt/intel/oneapi/ccl/2021.5.0/lib/cpu_gpu_dpcpp
                    -Wl,-rpath,/usr/x86_64-suse-linux/lib
                    -L/usr/x86_64-suse-linux/lib
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib/release
                    -Wl,-rpath,/opt/intel/oneapi/mpi/2021.5.0/lib
                    -lgfortran -lm -lgcc_s -lquadmath"
                    -Denable_parmetislib=FALSE
                    -DTPL_ENABLE_PARMETISLIB=FALSE
                    -DXSDK_ENABLE_Fortran=ON -Denable_tests=0
                    -Denable_examples=0 -DMPI_C_COMPILE_FLAGS:STRING=""
                    -DMPI_C_INCLUDE_PATH:STRING=""
                    -DMPI_C_HEADER_DIR:STRING=""
                    -DMPI_C_LIBRARIES:STRING=""</div>
                  <div><br>
                  </div>
                  <div><br>
                  </div>
                  <div><br>
                  </div>
                  <div><br>
                    <blockquote type="cite">
                      <div>On Dec 20, 2021, at 2:59 PM, Roland
                        Richter <<a href="mailto:roland.richter@ntnu.no" target="_blank">roland.richter@ntnu.no</a>>
                        wrote:</div>
                      <br>
                      <div>
                        <div>
                          <p>I introduced the changes from that
                            patch directly, without checking out. Is
                            that insufficient?</p>
                          <p>Regards,</p>
                          <p>Roland<br>
                          </p>
                          <div>Am 20.12.2021 um
                            20:38 schrieb Barry Smith:<br>
                          </div>
                          <blockquote type="cite">
                            <div><br>
                            </div>
                              Are you sure you have the correct PETSc
                            branch? From configure.log it has
                            <div><br>
                            </div>
                            <div>
                              <div>            Defined
                                "VERSION_GIT" to
                                ""v3.16.2-466-g959e1fce86""</div>
                              <div>            Defined
                                "VERSION_DATE_GIT" to ""2021-12-18
                                11:17:24 -0600""</div>
                              <div>            Defined
                                "VERSION_BRANCH_GIT" to ""master""</div>
                              <div><br>
                              </div>
                              <div>It should have
                                balay/slu-without-omp-3 for the branch.</div>
                              <div><br>
                              </div>
                              <div><br>
                              </div>
                              <div><br>
                                <blockquote type="cite">
                                  <div>On Dec 20, 2021, at
                                    10:50 AM, Roland Richter <<a href="mailto:roland.richter@ntnu.no" target="_blank">roland.richter@ntnu.no</a>>
                                    wrote:</div>
                                  <br>
                                  <div>
                                    <div>
                                      <p>In that case it fails
                                        with</p>
                                      <p><i>~/Downloads/git-files/petsc/mpich-complex-linux-gcc-demo/externalpackages/git.superlu_dist/SRC/cublas_utils.h:22:10:
                                          fatal error: cublas_v2.h: No
                                          such file or directory</i></p>
                                      <p>even though this
                                        header is available. I assume
                                        some header paths are not set
                                        correctly?</p>
                                      <p>Thanks,</p>
                                      <p>regards,</p>
                                      <p>Roland<br>
                                      </p>
                                      <div>Am
                                        20.12.21 um 16:29 schrieb Barry
                                        Smith:<br>
                                      </div>
                                      <blockquote type="cite">
                                        <div><br>
                                        </div>
                                          Please try the
                                        branch balay/slu-without-omp-3
                                         It is in MR <a href="https://gitlab.com/petsc/petsc/-/merge_requests/4635" target="_blank">https://gitlab.com/petsc/petsc/-/merge_requests/4635</a>
                                        <div><br>
                                        </div>
                                        <div><br>
                                          <div><br>
                                            <blockquote type="cite">
                                              <div>On Dec 20,
                                                2021, at 8:14 AM, Roland
                                                Richter <<a href="mailto:roland.richter@ntnu.no" target="_blank">roland.richter@ntnu.no</a>>
                                                wrote:</div>
                                              <br>
                                              <div>
                                                <div>
                                                  <p>Hei,</p>
                                                  <p>I tried to
                                                    combine CUDA with
                                                    superlu_dist in
                                                    petsc using the
                                                    following
                                                    configure-line:</p>
                                                  <p><i>./configure
PETSC_ARCH=mpich-complex-linux-gcc-demo
--CC=/opt/intel/oneapi/mpi/2021.5.0/bin/mpicc
                                                      --CXX=/opt/intel/oneapi/mpi/2021.5.0/bin/mpicxx
--FC=/opt/intel/oneapi/mpi/2021.5.0/bin/mpifc --CFLAGS="-mavx2
                                                      -march=native -O3"
                                                      --CXXFLAGS="-mavx2
                                                      -march=native -O3"
                                                      --FFLAGS="-mavx2
                                                      -march=native -O3"
--CUDAFLAGS=-allow-unsupported-compiler --CUDA-CXX=g++
                                                      --prefix=/opt/petsc
--with-blaslapack=1 --with-mpi=1 --with-scalar-type=complex
                                                      --download-suitesparse=1
                                                      --with-cuda
                                                      --with-debugging=0
                                                      --with-openmp
                                                      --download-superlu_dist
                                                      --force</i></p>
                                                  <p>but the
                                                    configure-step fails
                                                    with several errors
                                                    correlated with CUDA
                                                    and superlu_dist,
                                                    the first one being</p>
                                                  <p><i>cublas_utils.c:21:37:
                                                      error:
                                                      ‘CUDART_VERSION’
                                                      undeclared (first
                                                      use in this
                                                      function); did you
                                                      mean
                                                      ‘CUDA_VERSION’?</i><i><br>
                                                    </i><i>  
                                                      21 |    
                                                      printf("CUDA
                                                      version:   v
                                                      %d\n",CUDART_VERSION);</i><i><br>
                                                    </i><i>     
|                                     ^~~~~~~~~~~~~~</i><i><br>
                                                    </i><i>     
|                                     CUDA_VERSION</i></p>
                                                  <p>Compiling
                                                    superlu_dist
                                                    separately works,
                                                    though (including
                                                    CUDA).</p>
                                                  <p>Is there a
                                                    bug somewhere in the
                                                    configure-routine? I
                                                    attached the full
                                                    configure-log.<br>
                                                  </p>
                                                  <p>Thanks!</p>
                                                  <p>Regards,</p>
                                                  <p>Roland<br>
                                                  </p>
                                                </div>
                                                <span id="gmail-m_-6005507413301060485cid:F1E9B5DB-A92E-4C95-83FD-3F462C695FE9"><configure.log></span></div>
                                            </blockquote>
                                          </div>
                                          <br>
                                        </div>
                                      </blockquote>
                                    </div>
                                    <span id="gmail-m_-6005507413301060485cid:4A36F88D-58F8-40D3-A642-41C361B045C8"><configure.log></span></div>
                                </blockquote>
                              </div>
                              <br>
                            </div>
                          </blockquote>
                        </div>
                      </div>
                    </blockquote>
                  </div>
                  <br>
                </blockquote>
              </blockquote>
            </div>
            <span id="gmail-m_-6005507413301060485cid:16632CC5-01A0-4F67-AA92-6F8BEBEE763B"><configure.log></span></div>
        </blockquote>
      </div>
      <br>
    </blockquote>
  </div>
</blockquote></div>