<div dir="ltr"><div dir="ltr"><div>Hello,</div><div><br></div><div>I am trying to build
 CP2K 2022.2 for NVIDIA GPUs. I've built  ELPA with GPU support (no OpenMP, yes MPI), and COSMA 2.5 with GPU support too. The error I am getting is 
as follows:</div><div><br></div><div>mpifort -c -D__MKL -D__FFTW3 
-D__LIBXSMM  -D__LIBINT -D__parallel -D__SCALAPACK -D__HAS_ISO_C_BINDING
 -D__ACC -D__DBCSR_ACC -D__PW_CUDA -D__GRID_CUDA -D__OFFLOAD_PROFILING 
-D__HAS_PATCHED_CUFFT_70 -D__COSMA -D__ELPA_NVIDIA_GPU  -D__LIBVORI 
-D__SPGLIB -O3 -fopenmp -ffree-form -ffree-line-length-none 
-ftree-vectorize -funroll-loops -std=f2008 -I/opt/intel/compilers_and_libraries_2019.5.281/linux/mkl/include  -I/home/COSMA/include -I/home/libxsmm-1.17.0/include   -I/home/elpa-2021.11.002/include/elpa-2021.11.002/modules -I/home/elpa-2021.11.002/include/elpa-2021.11.002/elpa -I/home/elpa-2021.11.002/include/elpa-2021.11.002/src/helpers -I/home/fftw-3.3.10/include -I/home/libint-v2.6.0-cp2k-lmax-4/include -I/cuda/cuda-11.2/include -m64 -I/opt/intel/compilers_and_libraries_2019.5.281/linux/mkl/include -D__COMPILE_ARCH="\"nvidiaGPU\""  -D__COMPILE_REVISION="\"git:a95ec40\"" -D__DATA_DIR="\"/home/cp2k-2022.2/data\"" -D__SHORT_FILE__="\"xc/xc_input_constants.F\"" -I'/home/cp2k-2022.2/src/xc/' -I'/home/cp2k-2022.2/obj/nvidiaGPU/psmp/exts/dbcsr' xc_input_constants.F90 <br></div><div><br></div><div>/home/cp2k-2022.2/tools/build_utils/fypp -n /home/cp2k-2022.2/src/shg_int/constants_operator.F constants_operator.F90<br>cmp: /home/cp2k-2022.2/obj/git-ref: No such file or directory<br>/home/cp2k-2022.2/src/pw/gpu/pw_gpu_kernels.h(22): error: identifier "offloadStream_t" is undefined</div><div><br></div><div>and here you can see my arch file:</div><div><br></div><div><br></div><div>NVCC     = ${CUDA_HOME}/bin/nvcc</div>GPUVER   = V100<br><br>CXX      = mpicxx<br>CC       = mpicc<br>FC       = mpifort<br>LD       = mpifort<br>AR       = ar -r<br><br>OFFLOAD_CC = nvcc<br>OFFLOAD_FLAGS = -O3 -g -w --std=c++11<br><br>OPTFLAGS += -fopenmp -O3 -march=native<br><br>DFLAGS   += -D__MKL -D__FFTW3 -D__LIBXSMM  -D__LIBINT \<br>       -D__parallel -D__SCALAPACK -D__HAS_ISO_C_BINDING<br>     <br>DFLAGS += -D__ACC -D__DBCSR_ACC -D__PW_CUDA -D__GRID_CUDA -D__OFFLOAD_PROFILING<br>DFLAGS += -D__HAS_PATCHED_CUFFT_70 -D__COSMA -D__ELPA_NVIDIA_GPU <br><br>DFLAGS+= -D__LIBVORI -D__SPGLIB<br><br># C++ flags<br>CXXFLAGS = -O3 $(OPTFLAGS) -g -I${CUDA_HOME}/include -std=c++11<br><br># NVIDIA <br>NVFLAGS  = $(DFLAGS) $(OPTFLAGS) -g -O3 -arch sm_70 -Xcompiler='-fopenmp' --std=c++11<br><br># Fortran flags<br>FCFLAGS  = $(DFLAGS) -O3 -fopenmp -ffree-form -ffree-line-length-none<br>FCFLAGS    += -ftree-vectorize -funroll-loops -std=f2008<br>INC_FLAGS    = -I${MKLROOT}/include <br>INC_FLAGS    += -I$(COSMA)/include<br>INC_FLAGS    += -I$(LIBXSMM)/include  <br>INC_FLAGS    += -I$(INCELPA)/modules -I$(INCELPA)/elpa -I$(INCELPA)/src/helpers<br>INC_FLAGS    += -I$(LIBFFTW3)/include<br>INC_FLAGS    += -I$(LIBINT)/include<br>INC_FLAGS    += -I${CUDA_HOME}/include<br>INC_FLAGS    += -m64 -I${MKLROOT}/include<br><br># C++ flags<br>CXXFLAGS += $(INC_FLAGS)<br><br>FCFLAGS +=  $(INC_FLAGS)<br><br># C flags<br>CFLAGS      = $(DFLAGS) $(OPTFLAGS) $(INC_FLAGS)<br><br># Libraries<br>LDFLAGS  = $(FCFLAGS) <br>LDFLAGS_C = $(FCFLAGS)<br><br># ELPA libs<br>LIBS     = $(LIBELPA)/lib/libelpa.a<br># Libvori lib<br>LIBS     += $(LIBVORI)/lib/libvori.a<br># LIBXSMM libs<br>LIBS    += -L$(LIBXSMM)/lib -lxsmmf -lxsmm -lxsmmext  <br># LIBINT libs<br>LIBS    += -L$(LIBINT)/lib -lint2 -lstdc++<br># FFTW libs<br>LIBS    += -L$(LIBFFTW3)/lib -lfftw3 -lfftw3_threads<br>#SPGLIB<br>LIBS    += -L$(LIBSPG)/lib64 -lsymspg<br>#LIBXC<br>LIBS      += -L$(LIBXC)/lib -lxcf03 -lxc<br></div><div dir="ltr"><br></div><div dir="ltr"># Scalapack libs<br>LIBS    += -L/usr/lib64<br><br>LIBS    += -L$(MKLROOT)/lib/intel64 -Wl,--start-group -lmkl_scalapack_lp64 -lmkl_gf_lp64 -lmkl_sequential -lmkl_core -l<br>mkl_blacs_openmpi_lp64 -Wl,--end-group -lpthread -lz -ldl<br>  <br>#COSMA<br>LIBS    += -L$(COSMA_LIB)/ -lcosma_prefixed_pxgemm -lcosma -lcosta_prefixed_scalapack -lcosta<br><br>#CUDA<br>LIBS   += -L${CUDA_HOME}/lib64 -lstdc++ -lcudart -lnvrtc -lcuda -lcufft -lcublas -lnvToolsExt<br><br># Required due to memory leak that occurs if high optimisations are used<br>mp2_optimize_ri_basis.o: mp2_optimize_ri_basis.F<br>                       $(FC) -c $(subst O2,O0,$(FCFLAGS)) $<<div class="gmail-yj6qo"></div><div class="gmail-adL"><br></div><div class="gmail-adL">Could anyone help me with this please?</div><div class="gmail-adL"><br></div><div class="gmail-adL">Regards</div><div class="gmail-adL">LC<br></div></div></div>

<p></p>

-- <br />
You received this message because you are subscribed to the Google Groups "cp2k" group.<br />
To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:cp2k+unsubscribe@googlegroups.com">cp2k+unsubscribe@googlegroups.com</a>.<br />
To view this discussion on the web visit <a href="https://groups.google.com/d/msgid/cp2k/CAPm5di%3DFJHxW9C4kUYLVvpMiujZ9-oJ4W6JDRXjUUsQYG%2BYDpA%40mail.gmail.com?utm_medium=email&utm_source=footer">https://groups.google.com/d/msgid/cp2k/CAPm5di%3DFJHxW9C4kUYLVvpMiujZ9-oJ4W6JDRXjUUsQYG%2BYDpA%40mail.gmail.com</a>.<br />