Delete outdated arch files

This commit is contained in:
Matthias Krack 2024-08-16 16:05:57 +02:00
parent 42d1e6b4da
commit 68d95bacfe
2 changed files with 0 additions and 917 deletions

View file

@ -1,458 +0,0 @@
#!/bin/bash
#
# CP2K arch file for JURECA at FZ Juelich (2 AMD EPYC 7742 CPUs per node)
#
# Tested with: GNU 11.3.0, ParaStationMPI,
# ScaLAPACK 2.2.1, OpenBLAS 0.3.28,
# FFTW 3.3.10, COSMA 2.6.6,
# ELPA 2024.03.001, HDF5 1.14.2,
# LIBINT 2.6.0, LIBPEXSI 1.2.0,
# LIBXC 6.2.2, LIBVORI 220621,
# LIBXSMM 1.17, PLUMED 2.9.0,
# SIRIUS 7.5.2, SPGLIB 2.3.1
#
# Usage: Source this arch file and then run make as instructed.
# A full toolchain installation is performed as default.
# Replace or adapt the "module add" commands below if needed.
#
# Last update: 12.08.2024
#
# \
if [ "${0}" = "${BASH_SOURCE}" ]; then \
echo "ERROR: Script ${0##*/} must be sourced"; \
echo "Usage: source ${0##*/}"; \
exit 1; \
fi; \
this_file=${BASH_SOURCE##*/}; \
module purge; \
module add GCC/11.3.0; \
module add ParaStationMPI/5.8.0-1-mt; \
module list; \
module save cp2k_jureca_cpu_psmp; \
echo "To load the required modules in your batch job script, use:"; \
echo " module restore cp2k_jureca_cpu_psmp"; \
cd tools/toolchain; \
rm -rf build; \
./install_cp2k_toolchain.sh --enable-cuda=no --install-all -j32 --no-arch-files --with-gcc=system --with-mpich=system; \
cd ../..; \
printf "Sourcing ${PWD}/tools/toolchain/install/setup ... "; \
source ${PWD}/tools/toolchain/install/setup; \
printf "done\n"; \
echo "Check the output above for error messages and consistency!"; \
echo; \
echo "If everything is OK, you can build a CP2K production binary with"; \
echo " make -j ARCH=${this_file%.*} VERSION=${this_file##*.}"; \
echo; \
echo "Alternatively, you can add further checks, e.g. for regression testing, with"; \
echo " make -j ARCH=${this_file%.*} VERSION=${this_file##*.} DO_CHECKS=yes"; \
echo "or build CP2K as a library with"; \
echo " make -j ARCH=${this_file%.*} VERSION=${this_file##*.} libcp2k"; \
echo; \
return
# Set options
DO_CHECKS := no
SHARED := no
TARGET_CPU := native
USE_ACC := no
USE_COSMA := 2.6.6
USE_ELPA := 2024.03.001
USE_FFTW := 3.3.10
USE_HDF5 := 1.14.2
USE_LIBINT := 2.6.0
USE_LIBPEXSI := 1.2.0
USE_LIBTORCH := 1.12.1
USE_LIBVORI := 220621
USE_LIBXC := 6.2.2
USE_LIBXSMM := 1.17
USE_OPENBLAS := 0.3.28
USE_PLUMED := 2.9.0
USE_QUIP := 0.9.10
USE_SCALAPACK := 2.2.1
USE_SIRIUS := 7.5.2
USE_SPGLIB := 2.3.1
# Only needed for SIRIUS
LIBVDWXC_VER := 0.4.0
SPFFT_VER := 1.0.6
SPLA_VER := 1.5.5
# Only needed for LIBPEXSI
SCOTCH_VER := 6.0.0
SUPERLU_VER := 6.1.0
LMAX := 5
MAX_CONTR := 4
ifeq ($(USE_ACC), yes)
GPUVER := A100
OFFLOAD_TARGET := cuda
OFFLOAD_CC := nvcc
endif
CC := mpicc
CXX := mpicxx
FC := mpifort
LD := mpifort
AR := ar -r
# cc, CC, and ftn include already the proper -march flag
CFLAGS := -O2 -fPIC -fopenmp -fopenmp-simd -ftree-vectorize -funroll-loops -g -mtune=$(TARGET_CPU)
DFLAGS := -D__parallel
DFLAGS += -D__MAX_CONTR=$(strip $(MAX_CONTR))
INSTALL_PATH := $(PWD)/tools/toolchain/install
ifeq ($(SHARED), yes)
LD_SHARED := $(FC) -shared
CFLAGS += -fPIC
LDFLAGS := -Wl,--enable-new-dtags
CP2K_LIB := $(PWD)/lib/$(ARCH)/$(ONEVERSION)
LDFLAGS += -Wl,-rpath=$(CP2K_LIB)
LDFLAGS += -Wl,-rpath=$(CP2K_LIB)/exts/dbcsr
endif
# Settings for regression testing
ifeq ($(DO_CHECKS), yes)
DFLAGS += -D__CHECK_DIAG
# CFLAGS_DEBUG := -fsanitize=address
CFLAGS_DEBUG := -fsanitize=leak
# FCFLAGS_DEBUG := -fcheck=bounds,do,recursion,pointer
FCFLAGS_DEBUG += -fcheck=all,no-array-temps
FCFLAGS_DEBUG += -ffpe-trap=invalid,overflow,zero
FCFLAGS_DEBUG += -fimplicit-none
FCFLAGS_DEBUG += -finit-derived
FCFLAGS_DEBUG += -finit-real=snan
FCFLAGS_DEBUG += -finit-integer=-42
FCFLAGS_DEBUG += -finline-matmul-limit=0
WFLAGS := -Werror=aliasing
WFLAGS += -Werror=ampersand
WFLAGS += -Werror=c-binding-type
WFLAGS += -Werror=conversion
WFLAGS += -Werror=intrinsic-shadow
WFLAGS += -Werror=intrinsics-std
WFLAGS += -Werror=line-truncation
WFLAGS += -Wrealloc-lhs
WFLAGS += -Werror=tabs
WFLAGS += -Werror=target-lifetime
WFLAGS += -Werror=underflow
WFLAGS += -Werror=unused-but-set-variable
WFLAGS += -Werror=unused-dummy-argument
WFLAGS += -Werror=unused-variable
endif
ifeq ($(USE_ACC), yes)
DFLAGS += -D__DBCSR_ACC
DFLAGS += -D__OFFLOAD_CUDA
# Possibly no performance gain with PW_CUDA currently
DFLAGS += -D__NO_OFFLOAD_PW
endif
ifneq ($(USE_PLUMED),)
USE_PLUMED := $(strip $(USE_PLUMED))
PLUMED_LIB := $(INSTALL_PATH)/plumed-$(USE_PLUMED)/lib
DFLAGS += -D__PLUMED2
USE_GSL := 2.7
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(PLUMED_LIB) -L$(PLUMED_LIB) -lplumed -lplumedKernel
else
LIBS += $(PLUMED_LIB)/libplumed.a
endif
endif
ifneq ($(USE_ELPA),)
USE_ELPA := $(strip $(USE_ELPA))
ifeq ($(USE_ACC), yes)
TARGET := nvidia
DFLAGS += -D__ELPA_NVIDIA_GPU
else
TARGET := cpu
endif
ELPA_INC := $(INSTALL_PATH)/elpa-$(USE_ELPA)/$(TARGET)/include/elpa_openmp-$(USE_ELPA)
ELPA_LIB := $(INSTALL_PATH)/elpa-$(USE_ELPA)/$(TARGET)/lib
CFLAGS += -I$(ELPA_INC)/elpa -I$(ELPA_INC)/modules
DFLAGS += -D__ELPA
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(ELPA_LIB) -L$(ELPA_LIB) -lelpa_openmp
else
LIBS += $(ELPA_LIB)/libelpa_openmp.a
endif
endif
ifneq ($(USE_QUIP),)
USE_QUIP := $(strip $(USE_QUIP))
QUIP_INC := $(INSTALL_PATH)/quip-$(USE_QUIP)/include
QUIP_LIB := $(INSTALL_PATH)/quip-$(USE_QUIP)/lib
CFLAGS += -I$(QUIP_INC)
DFLAGS += -D__QUIP
LIBS += $(QUIP_LIB)/libquip_core.a
LIBS += $(QUIP_LIB)/libatoms.a
LIBS += $(QUIP_LIB)/libFoX_sax.a
LIBS += $(QUIP_LIB)/libFoX_common.a
LIBS += $(QUIP_LIB)/libFoX_utils.a
LIBS += $(QUIP_LIB)/libFoX_fsys.a
endif
ifneq ($(USE_LIBPEXSI),)
USE_LIBPEXSI := $(strip $(USE_LIBPEXSI))
SCOTCH_VER := $(strip $(SCOTCH_VER))
SUPERLU_VER := $(strip $(SUPERLU_VER))
LIBPEXSI_INC := $(INSTALL_PATH)/pexsi-$(USE_LIBPEXSI)/include
LIBPEXSI_LIB := $(INSTALL_PATH)/pexsi-$(USE_LIBPEXSI)/lib
SCOTCH_INC := $(INSTALL_PATH)/scotch-$(SCOTCH_VER)/include
SCOTCH_LIB := $(INSTALL_PATH)/scotch-$(SCOTCH_VER)/lib
SUPERLU_INC := $(INSTALL_PATH)/superlu_dist-$(SUPERLU_VER)/include
SUPERLU_LIB := $(INSTALL_PATH)/superlu_dist-$(SUPERLU_VER)/lib
CFLAGS += -I$(LIBPEXSI_INC) -I$(SCOTCH_INC) -I$(SUPERLU_INC)
DFLAGS += -D__LIBPEXSI
LIBS += $(LIBPEXSI_LIB)/libpexsi.a
LIBS += $(SUPERLU_LIB)/libsuperlu_dist.a
LIBS += $(SCOTCH_LIB)/libptscotchparmetis.a
LIBS += $(SCOTCH_LIB)/libptscotch.a
LIBS += $(SCOTCH_LIB)/libptscotcherr.a
LIBS += $(SCOTCH_LIB)/libscotchmetis.a
LIBS += $(SCOTCH_LIB)/libscotch.a
endif
ifneq ($(USE_LIBVORI),)
USE_LIBVORI := $(strip $(USE_LIBVORI))
LIBVORI_LIB := $(INSTALL_PATH)/libvori-$(USE_LIBVORI)/lib
DFLAGS += -D__LIBVORI
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBVORI_LIB) -L$(LIBVORI_LIB) -lvori
else
LIBS += $(LIBVORI_LIB)/libvori.a
endif
endif
ifneq ($(USE_LIBXC),)
USE_LIBXC := $(strip $(USE_LIBXC))
LIBXC_INC := $(INSTALL_PATH)/libxc-$(USE_LIBXC)/include
LIBXC_LIB := $(INSTALL_PATH)/libxc-$(USE_LIBXC)/lib
CFLAGS += -I$(LIBXC_INC)
DFLAGS += -D__LIBXC
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBXC_LIB) -L$(LIBXC_LIB) -lxcf03 -lxc
else
LIBS += $(LIBXC_LIB)/libxcf03.a
LIBS += $(LIBXC_LIB)/libxc.a
endif
endif
ifneq ($(USE_LIBINT),)
USE_LIBINT := $(strip $(USE_LIBINT))
LMAX := $(strip $(LMAX))
LIBINT_INC := $(INSTALL_PATH)/libint-v$(USE_LIBINT)-cp2k-lmax-$(LMAX)/include
LIBINT_LIB := $(INSTALL_PATH)/libint-v$(USE_LIBINT)-cp2k-lmax-$(LMAX)/lib
CFLAGS += -I$(LIBINT_INC)
DFLAGS += -D__LIBINT
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBINT_LIB) -L$(LIBINT_LIB) -lint2
else
LIBS += $(LIBINT_LIB)/libint2.a
endif
endif
ifneq ($(USE_SPGLIB),)
USE_SPGLIB := $(strip $(USE_SPGLIB))
SPGLIB_INC := $(INSTALL_PATH)/spglib-$(USE_SPGLIB)/include
SPGLIB_LIB := $(INSTALL_PATH)/spglib-$(USE_SPGLIB)/lib
CFLAGS += -I$(SPGLIB_INC)
DFLAGS += -D__SPGLIB
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(SPGLIB_LIB) -L$(SPGLIB_LIB) -lsymspg
else
LIBS += $(SPGLIB_LIB)/libsymspg.a
endif
endif
ifneq ($(USE_LIBXSMM),)
USE_LIBXSMM := $(strip $(USE_LIBXSMM))
LIBXSMM_INC := $(INSTALL_PATH)/libxsmm-$(USE_LIBXSMM)/include
LIBXSMM_LIB := $(INSTALL_PATH)/libxsmm-$(USE_LIBXSMM)/lib
CFLAGS += -I$(LIBXSMM_INC)
DFLAGS += -D__LIBXSMM
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBXSMM_LIB) -L$(LIBXSMM_LIB) -lxsmmf -lxsmm
else
LIBS += $(LIBXSMM_LIB)/libxsmmf.a
LIBS += $(LIBXSMM_LIB)/libxsmm.a
endif
endif
ifneq ($(USE_SIRIUS),)
USE_SIRIUS := $(strip $(USE_SIRIUS))
LIBVDWXC_VER := $(strip $(LIBVDWXC_VER))
LIBVDWXC_INC := $(INSTALL_PATH)/libvdwxc-$(LIBVDWXC_VER)/include
LIBVDWXC_LIB := $(INSTALL_PATH)/libvdwxc-$(LIBVDWXC_VER)/lib
SPFFT_VER := $(strip $(SPFFT_VER))
SPFFT_INC := $(INSTALL_PATH)/SpFFT-$(SPFFT_VER)/include
SPLA_VER := $(strip $(SPLA_VER))
SPLA_INC := $(INSTALL_PATH)/SpLA-$(SPLA_VER)/include/spla
ifeq ($(USE_ACC), yes)
DFLAGS += -D__OFFLOAD_GEMM
SPFFT_LIB := $(INSTALL_PATH)/SpFFT-$(SPFFT_VER)/lib/cuda
SPLA_LIB := $(INSTALL_PATH)/SpLA-$(SPLA_VER)/lib/cuda
SIRIUS_INC := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/include/cuda
SIRIUS_LIB := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/lib/cuda
else
SPFFT_LIB := $(INSTALL_PATH)/SpFFT-$(SPFFT_VER)/lib
SPLA_LIB := $(INSTALL_PATH)/SpLA-$(SPLA_VER)/lib
SIRIUS_INC := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/include
SIRIUS_LIB := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/lib
endif
CFLAGS += -I$(LIBVDWXC_INC)
CFLAGS += -I$(SPFFT_INC)
CFLAGS += -I$(SPLA_INC)
CFLAGS += -I$(SIRIUS_INC)
DFLAGS += -D__LIBVDWXC
DFLAGS += -D__SPFFT
DFLAGS += -D__SPLA
DFLAGS += -D__SIRIUS
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(SIRIUS_LIB) -L$(SIRIUS_LIB) -lsirius
LIBS += -Wl,-rpath=$(SPLA_LIB) -L$(SPLA_LIB) -lspla
LIBS += -Wl,-rpath=$(SPFFT_LIB) -L$(SPFFT_LIB) -lspfft
LIBS += -Wl,-rpath=$(LIBVDWXC_LIB) -L$(LIBVDWXC_LIB) -lvdwxc
else
LIBS += $(SIRIUS_LIB)/libsirius.a
LIBS += $(SPLA_LIB)/libspla.a
LIBS += $(SPFFT_LIB)/libspfft.a
LIBS += $(LIBVDWXC_LIB)/libvdwxc.a
endif
endif
ifneq ($(USE_HDF5),)
USE_HDF5 := $(strip $(USE_HDF5))
HDF5_INC := $(INSTALL_PATH)/hdf5-$(USE_HDF5)/include
HDF5_LIB := $(INSTALL_PATH)/hdf5-$(USE_HDF5)/lib
CFLAGS += -I$(HDF5_INC)
DFLAGS += -D__HDF5
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath,$(HDF5_LIB) -L$(HDF5_LIB) -lhdf5_fortran -lhdf5
else
LIBS += $(HDF5_LIB)/libhdf5_fortran.a
LIBS += $(HDF5_LIB)/libhdf5_f90cstub.a
LIBS += $(HDF5_LIB)/libhdf5.a
endif
endif
ifneq ($(USE_COSMA),)
USE_COSMA := $(strip $(USE_COSMA))
ifeq ($(USE_ACC), yes)
USE_COSMA := $(USE_COSMA)-cuda
endif
COSMA_INC := $(INSTALL_PATH)/COSMA-$(USE_COSMA)/include
COSMA_LIB := $(INSTALL_PATH)/COSMA-$(USE_COSMA)/lib
CFLAGS += -I$(COSMA_INC)
DFLAGS += -D__COSMA
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(COSMA_LIB) -L$(COSMA_LIB) -lcosma_prefixed_pxgemm -lcosma -lcosta
ifeq ($(USE_ACC), yes)
LIBS += -lTiled-MM
endif
else
LIBS += $(COSMA_LIB)/libcosma_prefixed_pxgemm.a
LIBS += $(COSMA_LIB)/libcosma.a
LIBS += $(COSMA_LIB)/libcosta.a
ifeq ($(USE_ACC), yes)
LIBS += $(COSMA_LIB)/libTiled-MM.a
endif
endif
endif
ifneq ($(USE_FFTW),)
USE_FFTW := $(strip $(USE_FFTW))
FFTW_INC := $(INSTALL_PATH)/fftw-$(USE_FFTW)/include
FFTW_LIB := $(INSTALL_PATH)/fftw-$(USE_FFTW)/lib
CFLAGS += -I$(FFTW_INC)
DFLAGS += -D__FFTW3
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(FFTW_LIB) -L$(FFTW_LIB) -lfftw3_mpi -lfftw3_omp -lfftw3
else
LIBS += $(FFTW_LIB)/libfftw3_mpi.a
LIBS += $(FFTW_LIB)/libfftw3_omp.a
LIBS += $(FFTW_LIB)/libfftw3.a
endif
endif
ifneq ($(USE_SCALAPACK),)
SCALAPACK_LIB := $(INSTALL_PATH)/scalapack-$(USE_SCALAPACK)/lib
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(SCALAPACK_LIB) -L$(SCALAPACK_LIB) -lscalapack
else
LIBS += $(SCALAPACK_LIB)/libscalapack.a
endif
endif
ifneq ($(USE_GSL),)
USE_GSL := $(strip $(USE_GSL))
GSL_INC := $(INSTALL_PATH)/gsl-$(USE_GSL)/include
GSL_LIB := $(INSTALL_PATH)/gsl-$(USE_GSL)/lib
CFLAGS += -I$(GSL_INC)
DFLAGS += -D__GSL
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(GSL_LIB) -L$(GSL_LIB) -lgsl
else
LIBS += $(GSL_LIB)/libgsl.a
endif
endif
ifneq ($(USE_OPENBLAS),)
USE_OPENBLAS := $(strip $(USE_OPENBLAS))
OPENBLAS_INC := $(INSTALL_PATH)/openblas-$(USE_OPENBLAS)/include
OPENBLAS_LIB := $(INSTALL_PATH)/openblas-$(USE_OPENBLAS)/lib
CFLAGS += -I$(OPENBLAS_INC)
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(OPENBLAS_LIB) -L$(OPENBLAS_LIB) -lopenblas
else
LIBS += $(OPENBLAS_LIB)/libopenblas.a
endif
endif
ifeq ($(shell [ $(shell ldd --version | head -n 1 | tr -s '.' '\n' | tail -n 1) -ge 27 ] && echo yes), yes)
ifneq ($(USE_LIBTORCH),)
USE_LIBTORCH := $(strip $(USE_LIBTORCH))
LIBTORCH_INC := $(INSTALL_PATH)/libtorch-$(USE_LIBTORCH)/include
LIBTORCH_LIB := $(INSTALL_PATH)/libtorch-$(USE_LIBTORCH)/lib
CFLAGS += -I$(LIBTORCH_INC)
DFLAGS += -D__LIBTORCH
ifeq ($(USE_ACC), yes)
LIBS += -Wl,-rpath=$(LIBTORCH_LIB) -L$(LIBTORCH_LIB) -lc10 -ltorch_cpu -ltorch
# LIBS += -Wl,-rpath=$(LIBTORCH_LIB) -L$(LIBTORCH_LIB) -lc10 -lc10_cuda -ltorch_cpu -ltorch_cuda -ltorch
else
LIBS += -Wl,-rpath=$(LIBTORCH_LIB) -L$(LIBTORCH_LIB) -lc10 -ltorch_cpu -ltorch
endif
endif
endif
CFLAGS += $(DFLAGS)
FCFLAGS := $(CFLAGS)
ifeq ($(shell [ $(shell gcc -dumpversion | cut -d. -f1) -gt 9 ] && echo yes), yes)
FCFLAGS += -fallow-argument-mismatch
endif
FCFLAGS += -fbacktrace
FCFLAGS += -ffree-form
FCFLAGS += -ffree-line-length-none
FCFLAGS += -fno-omit-frame-pointer
FCFLAGS += -std=f2008
ifneq ($(CUDA_HOME),)
CFLAGS += -I$(CUDA_HOME)/include
FCFLAGS += -I$(CUDA_HOME)/include
CUDA_LIB := $(CUDA_HOME)/lib64
LDFLAGS += $(FCFLAGS) -L$(CUDA_LIB) -Wl,-rpath=$(CUDA_LIB)
else
LDFLAGS += $(FCFLAGS)
endif
CFLAGS += $(CFLAGS) -std=c11
CXXFLAGS := $(CFLAGS) -std=c++14
ifeq ($(USE_ACC), yes)
OFFLOAD_FLAGS := $(DFLAGS) -O3 -Xcompiler='-fopenmp' -allow-unsupported-compiler -arch sm_80 -g --std=c++11
LIBS += -lcusolver -lcudart -lnvrtc -lcuda -lcufft -lcublas -lrt
endif
LIBS += -lz -ldl -lpthread -lstdc++
# End

View file

@ -1,459 +0,0 @@
#!/bin/bash
#
# CP2K arch file for JURECA at FZ Juelich (2 AMD EPYC 7742 CPUs and 4 A100 GPUs per node)
#
# Tested with: GNU 11.3.0, ParaStationMPI,
# ScaLAPACK 2.2.1, OpenBLAS 0.3.28,
# FFTW 3.3.10, COSMA 2.6.6,
# ELPA 2024.03.001, HDF5 1.14.2,
# LIBINT 2.6.0, LIBPEXSI 1.2.0,
# LIBXC 6.2.2, LIBVORI 220621,
# LIBXSMM 1.17, PLUMED 2.9.0,
# SIRIUS 7.5.2, SPGLIB 2.3.1
#
# Usage: Source this arch file and then run make as instructed.
# A full toolchain installation is performed as default.
# Replace or adapt the "module add" commands below if needed.
#
# Last update: 12.08.2024
#
# \
if [ "${0}" = "${BASH_SOURCE}" ]; then \
echo "ERROR: Script ${0##*/} must be sourced"; \
echo "Usage: source ${0##*/}"; \
exit 1; \
fi; \
this_file=${BASH_SOURCE##*/}; \
module purge; \
module add GCC/11.3.0; \
module add ParaStationMPI/5.8.0-1-mt; \
module add CUDA/11.7; \
module list; \
module save cp2k_jureca_gpu_psmp; \
echo "To load the required modules in your batch job script, use:"; \
echo " module restore cp2k_jureca_gpu_psmp"; \
cd tools/toolchain; \
rm -rf build; \
./install_cp2k_toolchain.sh --enable-cuda=yes --gpu-ver=A100 --install-all -j32 --no-arch-files --with-gcc=system --with-mpich=system; \
cd ../..; \
printf "Sourcing ${PWD}/tools/toolchain/install/setup ... "; \
source ${PWD}/tools/toolchain/install/setup; \
printf "done\n"; \
echo "Check the output above for error messages and consistency!"; \
echo; \
echo "If everything is OK, you can build a CP2K production binary with"; \
echo " make -j ARCH=${this_file%.*} VERSION=${this_file##*.}"; \
echo; \
echo "Alternatively, you can add further checks, e.g. for regression testing, with"; \
echo " make -j ARCH=${this_file%.*} VERSION=${this_file##*.} DO_CHECKS=yes"; \
echo "or build CP2K as a library with"; \
echo " make -j ARCH=${this_file%.*} VERSION=${this_file##*.} libcp2k"; \
echo; \
return
# Set options
DO_CHECKS := no
SHARED := no
TARGET_CPU := native
USE_ACC := yes
USE_COSMA := 2.6.6
USE_ELPA := 2024.03.001
USE_FFTW := 3.3.10
USE_HDF5 := 1.14.2
USE_LIBINT := 2.6.0
USE_LIBPEXSI := 1.2.0
USE_LIBTORCH := 1.12.1
USE_LIBVORI := 220621
USE_LIBXC := 6.2.2
USE_LIBXSMM := 1.17
USE_OPENBLAS := 0.3.28
USE_PLUMED := 2.9.0
USE_QUIP := 0.9.10
USE_SCALAPACK := 2.2.1
USE_SIRIUS := 7.5.2
USE_SPGLIB := 2.3.1
# Only needed for SIRIUS
LIBVDWXC_VER := 0.4.0
SPFFT_VER := 1.0.6
SPLA_VER := 1.5.5
# Only needed for LIBPEXSI
SCOTCH_VER := 6.0.0
SUPERLU_VER := 6.1.0
LMAX := 5
MAX_CONTR := 4
ifeq ($(USE_ACC), yes)
GPUVER := A100
OFFLOAD_TARGET := cuda
OFFLOAD_CC := nvcc
endif
CC := mpicc
CXX := mpicxx
FC := mpifort
LD := mpifort
AR := ar -r
# cc, CC, and ftn include already the proper -march flag
CFLAGS := -O2 -fPIC -fopenmp -fopenmp-simd -ftree-vectorize -funroll-loops -g -mtune=$(TARGET_CPU)
DFLAGS := -D__parallel
DFLAGS += -D__MAX_CONTR=$(strip $(MAX_CONTR))
INSTALL_PATH := $(PWD)/tools/toolchain/install
ifeq ($(SHARED), yes)
LD_SHARED := $(FC) -shared
CFLAGS += -fPIC
LDFLAGS := -Wl,--enable-new-dtags
CP2K_LIB := $(PWD)/lib/$(ARCH)/$(ONEVERSION)
LDFLAGS += -Wl,-rpath=$(CP2K_LIB)
LDFLAGS += -Wl,-rpath=$(CP2K_LIB)/exts/dbcsr
endif
# Settings for regression testing
ifeq ($(DO_CHECKS), yes)
DFLAGS += -D__CHECK_DIAG
# CFLAGS_DEBUG := -fsanitize=address
CFLAGS_DEBUG := -fsanitize=leak
# FCFLAGS_DEBUG := -fcheck=bounds,do,recursion,pointer
FCFLAGS_DEBUG += -fcheck=all,no-array-temps
FCFLAGS_DEBUG += -ffpe-trap=invalid,overflow,zero
FCFLAGS_DEBUG += -fimplicit-none
FCFLAGS_DEBUG += -finit-derived
FCFLAGS_DEBUG += -finit-real=snan
FCFLAGS_DEBUG += -finit-integer=-42
FCFLAGS_DEBUG += -finline-matmul-limit=0
WFLAGS := -Werror=aliasing
WFLAGS += -Werror=ampersand
WFLAGS += -Werror=c-binding-type
WFLAGS += -Werror=conversion
WFLAGS += -Werror=intrinsic-shadow
WFLAGS += -Werror=intrinsics-std
WFLAGS += -Werror=line-truncation
WFLAGS += -Wrealloc-lhs
WFLAGS += -Werror=tabs
WFLAGS += -Werror=target-lifetime
WFLAGS += -Werror=underflow
WFLAGS += -Werror=unused-but-set-variable
WFLAGS += -Werror=unused-dummy-argument
WFLAGS += -Werror=unused-variable
endif
ifeq ($(USE_ACC), yes)
DFLAGS += -D__DBCSR_ACC
DFLAGS += -D__OFFLOAD_CUDA
# Possibly no performance gain with PW_CUDA currently
DFLAGS += -D__NO_OFFLOAD_PW
endif
ifneq ($(USE_PLUMED),)
USE_PLUMED := $(strip $(USE_PLUMED))
PLUMED_LIB := $(INSTALL_PATH)/plumed-$(USE_PLUMED)/lib
DFLAGS += -D__PLUMED2
USE_GSL := 2.7
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(PLUMED_LIB) -L$(PLUMED_LIB) -lplumed -lplumedKernel
else
LIBS += $(PLUMED_LIB)/libplumed.a
endif
endif
ifneq ($(USE_ELPA),)
USE_ELPA := $(strip $(USE_ELPA))
ifeq ($(USE_ACC), yes)
TARGET := nvidia
DFLAGS += -D__ELPA_NVIDIA_GPU
else
TARGET := cpu
endif
ELPA_INC := $(INSTALL_PATH)/elpa-$(USE_ELPA)/$(TARGET)/include/elpa_openmp-$(USE_ELPA)
ELPA_LIB := $(INSTALL_PATH)/elpa-$(USE_ELPA)/$(TARGET)/lib
CFLAGS += -I$(ELPA_INC)/elpa -I$(ELPA_INC)/modules
DFLAGS += -D__ELPA
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(ELPA_LIB) -L$(ELPA_LIB) -lelpa_openmp
else
LIBS += $(ELPA_LIB)/libelpa_openmp.a
endif
endif
ifneq ($(USE_QUIP),)
USE_QUIP := $(strip $(USE_QUIP))
QUIP_INC := $(INSTALL_PATH)/quip-$(USE_QUIP)/include
QUIP_LIB := $(INSTALL_PATH)/quip-$(USE_QUIP)/lib
CFLAGS += -I$(QUIP_INC)
DFLAGS += -D__QUIP
LIBS += $(QUIP_LIB)/libquip_core.a
LIBS += $(QUIP_LIB)/libatoms.a
LIBS += $(QUIP_LIB)/libFoX_sax.a
LIBS += $(QUIP_LIB)/libFoX_common.a
LIBS += $(QUIP_LIB)/libFoX_utils.a
LIBS += $(QUIP_LIB)/libFoX_fsys.a
endif
ifneq ($(USE_LIBPEXSI),)
USE_LIBPEXSI := $(strip $(USE_LIBPEXSI))
SCOTCH_VER := $(strip $(SCOTCH_VER))
SUPERLU_VER := $(strip $(SUPERLU_VER))
LIBPEXSI_INC := $(INSTALL_PATH)/pexsi-$(USE_LIBPEXSI)/include
LIBPEXSI_LIB := $(INSTALL_PATH)/pexsi-$(USE_LIBPEXSI)/lib
SCOTCH_INC := $(INSTALL_PATH)/scotch-$(SCOTCH_VER)/include
SCOTCH_LIB := $(INSTALL_PATH)/scotch-$(SCOTCH_VER)/lib
SUPERLU_INC := $(INSTALL_PATH)/superlu_dist-$(SUPERLU_VER)/include
SUPERLU_LIB := $(INSTALL_PATH)/superlu_dist-$(SUPERLU_VER)/lib
CFLAGS += -I$(LIBPEXSI_INC) -I$(SCOTCH_INC) -I$(SUPERLU_INC)
DFLAGS += -D__LIBPEXSI
LIBS += $(LIBPEXSI_LIB)/libpexsi.a
LIBS += $(SUPERLU_LIB)/libsuperlu_dist.a
LIBS += $(SCOTCH_LIB)/libptscotchparmetis.a
LIBS += $(SCOTCH_LIB)/libptscotch.a
LIBS += $(SCOTCH_LIB)/libptscotcherr.a
LIBS += $(SCOTCH_LIB)/libscotchmetis.a
LIBS += $(SCOTCH_LIB)/libscotch.a
endif
ifneq ($(USE_LIBVORI),)
USE_LIBVORI := $(strip $(USE_LIBVORI))
LIBVORI_LIB := $(INSTALL_PATH)/libvori-$(USE_LIBVORI)/lib
DFLAGS += -D__LIBVORI
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBVORI_LIB) -L$(LIBVORI_LIB) -lvori
else
LIBS += $(LIBVORI_LIB)/libvori.a
endif
endif
ifneq ($(USE_LIBXC),)
USE_LIBXC := $(strip $(USE_LIBXC))
LIBXC_INC := $(INSTALL_PATH)/libxc-$(USE_LIBXC)/include
LIBXC_LIB := $(INSTALL_PATH)/libxc-$(USE_LIBXC)/lib
CFLAGS += -I$(LIBXC_INC)
DFLAGS += -D__LIBXC
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBXC_LIB) -L$(LIBXC_LIB) -lxcf03 -lxc
else
LIBS += $(LIBXC_LIB)/libxcf03.a
LIBS += $(LIBXC_LIB)/libxc.a
endif
endif
ifneq ($(USE_LIBINT),)
USE_LIBINT := $(strip $(USE_LIBINT))
LMAX := $(strip $(LMAX))
LIBINT_INC := $(INSTALL_PATH)/libint-v$(USE_LIBINT)-cp2k-lmax-$(LMAX)/include
LIBINT_LIB := $(INSTALL_PATH)/libint-v$(USE_LIBINT)-cp2k-lmax-$(LMAX)/lib
CFLAGS += -I$(LIBINT_INC)
DFLAGS += -D__LIBINT
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBINT_LIB) -L$(LIBINT_LIB) -lint2
else
LIBS += $(LIBINT_LIB)/libint2.a
endif
endif
ifneq ($(USE_SPGLIB),)
USE_SPGLIB := $(strip $(USE_SPGLIB))
SPGLIB_INC := $(INSTALL_PATH)/spglib-$(USE_SPGLIB)/include
SPGLIB_LIB := $(INSTALL_PATH)/spglib-$(USE_SPGLIB)/lib
CFLAGS += -I$(SPGLIB_INC)
DFLAGS += -D__SPGLIB
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(SPGLIB_LIB) -L$(SPGLIB_LIB) -lsymspg
else
LIBS += $(SPGLIB_LIB)/libsymspg.a
endif
endif
ifneq ($(USE_LIBXSMM),)
USE_LIBXSMM := $(strip $(USE_LIBXSMM))
LIBXSMM_INC := $(INSTALL_PATH)/libxsmm-$(USE_LIBXSMM)/include
LIBXSMM_LIB := $(INSTALL_PATH)/libxsmm-$(USE_LIBXSMM)/lib
CFLAGS += -I$(LIBXSMM_INC)
DFLAGS += -D__LIBXSMM
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(LIBXSMM_LIB) -L$(LIBXSMM_LIB) -lxsmmf -lxsmm
else
LIBS += $(LIBXSMM_LIB)/libxsmmf.a
LIBS += $(LIBXSMM_LIB)/libxsmm.a
endif
endif
ifneq ($(USE_SIRIUS),)
USE_SIRIUS := $(strip $(USE_SIRIUS))
LIBVDWXC_VER := $(strip $(LIBVDWXC_VER))
LIBVDWXC_INC := $(INSTALL_PATH)/libvdwxc-$(LIBVDWXC_VER)/include
LIBVDWXC_LIB := $(INSTALL_PATH)/libvdwxc-$(LIBVDWXC_VER)/lib
SPFFT_VER := $(strip $(SPFFT_VER))
SPFFT_INC := $(INSTALL_PATH)/SpFFT-$(SPFFT_VER)/include
SPLA_VER := $(strip $(SPLA_VER))
SPLA_INC := $(INSTALL_PATH)/SpLA-$(SPLA_VER)/include/spla
ifeq ($(USE_ACC), yes)
DFLAGS += -D__OFFLOAD_GEMM
SPFFT_LIB := $(INSTALL_PATH)/SpFFT-$(SPFFT_VER)/lib/cuda
SPLA_LIB := $(INSTALL_PATH)/SpLA-$(SPLA_VER)/lib/cuda
SIRIUS_INC := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/include/cuda
SIRIUS_LIB := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/lib/cuda
else
SPFFT_LIB := $(INSTALL_PATH)/SpFFT-$(SPFFT_VER)/lib
SPLA_LIB := $(INSTALL_PATH)/SpLA-$(SPLA_VER)/lib
SIRIUS_INC := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/include
SIRIUS_LIB := $(INSTALL_PATH)/sirius-$(USE_SIRIUS)/lib
endif
CFLAGS += -I$(LIBVDWXC_INC)
CFLAGS += -I$(SPFFT_INC)
CFLAGS += -I$(SPLA_INC)
CFLAGS += -I$(SIRIUS_INC)
DFLAGS += -D__LIBVDWXC
DFLAGS += -D__SPFFT
DFLAGS += -D__SPLA
DFLAGS += -D__SIRIUS
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(SIRIUS_LIB) -L$(SIRIUS_LIB) -lsirius
LIBS += -Wl,-rpath=$(SPLA_LIB) -L$(SPLA_LIB) -lspla
LIBS += -Wl,-rpath=$(SPFFT_LIB) -L$(SPFFT_LIB) -lspfft
LIBS += -Wl,-rpath=$(LIBVDWXC_LIB) -L$(LIBVDWXC_LIB) -lvdwxc
else
LIBS += $(SIRIUS_LIB)/libsirius.a
LIBS += $(SPLA_LIB)/libspla.a
LIBS += $(SPFFT_LIB)/libspfft.a
LIBS += $(LIBVDWXC_LIB)/libvdwxc.a
endif
endif
ifneq ($(USE_HDF5),)
USE_HDF5 := $(strip $(USE_HDF5))
HDF5_INC := $(INSTALL_PATH)/hdf5-$(USE_HDF5)/include
HDF5_LIB := $(INSTALL_PATH)/hdf5-$(USE_HDF5)/lib
CFLAGS += -I$(HDF5_INC)
DFLAGS += -D__HDF5
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath,$(HDF5_LIB) -L$(HDF5_LIB) -lhdf5_fortran -lhdf5
else
LIBS += $(HDF5_LIB)/libhdf5_fortran.a
LIBS += $(HDF5_LIB)/libhdf5_f90cstub.a
LIBS += $(HDF5_LIB)/libhdf5.a
endif
endif
ifneq ($(USE_COSMA),)
USE_COSMA := $(strip $(USE_COSMA))
ifeq ($(USE_ACC), yes)
USE_COSMA := $(USE_COSMA)-cuda
endif
COSMA_INC := $(INSTALL_PATH)/COSMA-$(USE_COSMA)/include
COSMA_LIB := $(INSTALL_PATH)/COSMA-$(USE_COSMA)/lib
CFLAGS += -I$(COSMA_INC)
DFLAGS += -D__COSMA
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(COSMA_LIB) -L$(COSMA_LIB) -lcosma_prefixed_pxgemm -lcosma -lcosta
ifeq ($(USE_ACC), yes)
LIBS += -lTiled-MM
endif
else
LIBS += $(COSMA_LIB)/libcosma_prefixed_pxgemm.a
LIBS += $(COSMA_LIB)/libcosma.a
LIBS += $(COSMA_LIB)/libcosta.a
ifeq ($(USE_ACC), yes)
LIBS += $(COSMA_LIB)/libTiled-MM.a
endif
endif
endif
ifneq ($(USE_FFTW),)
USE_FFTW := $(strip $(USE_FFTW))
FFTW_INC := $(INSTALL_PATH)/fftw-$(USE_FFTW)/include
FFTW_LIB := $(INSTALL_PATH)/fftw-$(USE_FFTW)/lib
CFLAGS += -I$(FFTW_INC)
DFLAGS += -D__FFTW3
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(FFTW_LIB) -L$(FFTW_LIB) -lfftw3_mpi -lfftw3_omp -lfftw3
else
LIBS += $(FFTW_LIB)/libfftw3_mpi.a
LIBS += $(FFTW_LIB)/libfftw3_omp.a
LIBS += $(FFTW_LIB)/libfftw3.a
endif
endif
ifneq ($(USE_SCALAPACK),)
SCALAPACK_LIB := $(INSTALL_PATH)/scalapack-$(USE_SCALAPACK)/lib
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(SCALAPACK_LIB) -L$(SCALAPACK_LIB) -lscalapack
else
LIBS += $(SCALAPACK_LIB)/libscalapack.a
endif
endif
ifneq ($(USE_GSL),)
USE_GSL := $(strip $(USE_GSL))
GSL_INC := $(INSTALL_PATH)/gsl-$(USE_GSL)/include
GSL_LIB := $(INSTALL_PATH)/gsl-$(USE_GSL)/lib
CFLAGS += -I$(GSL_INC)
DFLAGS += -D__GSL
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(GSL_LIB) -L$(GSL_LIB) -lgsl
else
LIBS += $(GSL_LIB)/libgsl.a
endif
endif
ifneq ($(USE_OPENBLAS),)
USE_OPENBLAS := $(strip $(USE_OPENBLAS))
OPENBLAS_INC := $(INSTALL_PATH)/openblas-$(USE_OPENBLAS)/include
OPENBLAS_LIB := $(INSTALL_PATH)/openblas-$(USE_OPENBLAS)/lib
CFLAGS += -I$(OPENBLAS_INC)
ifeq ($(SHARED), yes)
LIBS += -Wl,-rpath=$(OPENBLAS_LIB) -L$(OPENBLAS_LIB) -lopenblas
else
LIBS += $(OPENBLAS_LIB)/libopenblas.a
endif
endif
ifeq ($(shell [ $(shell ldd --version | head -n 1 | tr -s '.' '\n' | tail -n 1) -ge 27 ] && echo yes), yes)
ifneq ($(USE_LIBTORCH),)
USE_LIBTORCH := $(strip $(USE_LIBTORCH))
LIBTORCH_INC := $(INSTALL_PATH)/libtorch-$(USE_LIBTORCH)/include
LIBTORCH_LIB := $(INSTALL_PATH)/libtorch-$(USE_LIBTORCH)/lib
CFLAGS += -I$(LIBTORCH_INC)
DFLAGS += -D__LIBTORCH
ifeq ($(USE_ACC), yes)
LIBS += -Wl,-rpath=$(LIBTORCH_LIB) -L$(LIBTORCH_LIB) -lc10 -ltorch_cpu -ltorch
# LIBS += -Wl,-rpath=$(LIBTORCH_LIB) -L$(LIBTORCH_LIB) -lc10 -lc10_cuda -ltorch_cpu -ltorch_cuda -ltorch
else
LIBS += -Wl,-rpath=$(LIBTORCH_LIB) -L$(LIBTORCH_LIB) -lc10 -ltorch_cpu -ltorch
endif
endif
endif
CFLAGS += $(DFLAGS)
FCFLAGS := $(CFLAGS)
ifeq ($(shell [ $(shell gcc -dumpversion | cut -d. -f1) -gt 9 ] && echo yes), yes)
FCFLAGS += -fallow-argument-mismatch
endif
FCFLAGS += -fbacktrace
FCFLAGS += -ffree-form
FCFLAGS += -ffree-line-length-none
FCFLAGS += -fno-omit-frame-pointer
FCFLAGS += -std=f2008
ifneq ($(CUDA_HOME),)
CFLAGS += -I$(CUDA_HOME)/include
FCFLAGS += -I$(CUDA_HOME)/include
CUDA_LIB := $(CUDA_HOME)/lib64
LDFLAGS += $(FCFLAGS) -L$(CUDA_LIB) -Wl,-rpath=$(CUDA_LIB)
else
LDFLAGS += $(FCFLAGS)
endif
CFLAGS += $(CFLAGS) -std=c11
CXXFLAGS := $(CFLAGS) -std=c++14
ifeq ($(USE_ACC), yes)
OFFLOAD_FLAGS := $(DFLAGS) -O3 -Xcompiler='-fopenmp' -allow-unsupported-compiler -arch sm_80 -g --std=c++11
LIBS += -lcusolver -lcudart -lnvrtc -lcuda -lcufft -lcublas -lrt
endif
LIBS += -lz -ldl -lpthread -lstdc++
# End