Arch Linux User Repository

At the moment build fails:

…
cat util_module_avail_F.end >> ../util/util_module_avail.F
make[1]: Leaving directory '~/.cache/yay/nwchem/src/nwchem-7.2.2-release/src/config'
bash: line 1: ~/.cache/yay/nwchem/src/nwchem-7.2.2-release/src/tools/install/bin/ga-config: No such file or directory
bash: line 1: ~/.cache/yay/nwchem/src/nwchem-7.2.2-release/src/tools/install/bin/ga-config: No such file or directory
cat: ~/.cache/yay/nwchem/src/nwchem-7.2.2-release/src/tools/build/global/src/ga-wapi.h: No such file or directory
cat: ~/.cache/yay/nwchem/src/nwchem-7.2.2-release/src/tools/build/global/src/ga-wapi.h: No such file or directory
bash: line 1: python3.11-config: command not found
(standard_in) 1: syntax error
(standard_in) 1: syntax error
GNUmakefile:27: warning: overriding recipe for target 'sngl_to_dbl'
../config/makelib.h:300: warning: ignoring old recipe for target 'sngl_to_dbl'
…
config/makefile.h:3290: *** .  Stop.
==> ERROR: A failure occurred in build().
    Aborting...
 -> error making: nwchem-exit status 4
 -> Failed to install the following packages. Manual intervention is required:
nwchem - exit status 4

It seems that this version of NWCHEM does not work after yay installs it. Sadly, I get this error:

[compyooter:1936134] pmix_mca_base_component_repository_open: unable to open mca_preg_compress: /usr/lib/openmpi/pmix/mca_preg_compress.so: undefined symbol: pmix_compress (ignored)
[compyooter:1936133] pmix_mca_base_component_repository_open: unable to open mca_preg_compress: /usr/lib/openmpi/pmix/mca_preg_compress.so: undefined symbol: pmix_compress (ignored)
--------------------------------------------------------------------------
The library attempted to open the following supporting CUDA libraries,
but each of them failed.  CUDA-aware support is disabled.
libcuda.so.1: cannot open shared object file: No such file or directory
libcuda.dylib: cannot open shared object file: No such file or directory
/usr/lib64/libcuda.so.1: cannot open shared object file: No such file or directory
/usr/lib64/libcuda.dylib: cannot open shared object file: No such file or directory
If you are not interested in CUDA-aware support, then run with
--mca opal_warn_on_missing_libcuda 0 to suppress this message.  If you are interested
in CUDA-aware support, then try setting LD_LIBRARY_PATH to the location
of libcuda.so.1 to get passed this issue.
--------------------------------------------------------------------------
[0] Received an Error in Communication: (1) there must be at least two ranks per node
--------------------------------------------------------------------------
MPI_ABORT was invoked on rank 0 in communicator MPI COMMUNICATOR 3 DUP FROM 0
with errorcode 1.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
--------------------------------------------------------------------------

Is there any way I can get NWCHEM to work? Thanks!

Please, disown the package if you don't want to maintain it anymore.

The USE_64TO32 option is for when the blas library only supports 32-bit integers, which the manual states is usually for precompiled vendor libraries. The absence will not cause a 32-bit build, but can cause incorrect results. The netlib blas library used in the Arch repositories should be 64-bit/"size of int: 8", but I may be working under a false assumption. However, the test-suite does complete successfully... The USE_64TO32 option will make the 64-bit version work regardless, but for those who definitely want a 64-bit blas implementation, I recommend OpenBLAS (which has a 64-bit interface option): https://aur.archlinux.org/packages/openblas/

As it is now, this build only works for 32-bit I think. If you could add support for 64-bit that would be great. All that has be changed I think is to add "make 64_to_32" in the PKGBUILD build() function. I use the following lines to build my version of NWChem. export NWCHEM_TOP=$PWD if [ `uname -m` == "x86_64" ]; then export NWCHEM_TARGET=LINUX64 export USE_64TO32=y else export NWCHEM_TARGET=LINUX fi export USE_MPI=y export BLASOPT="-lblas -lpthread -lrt" cd src make nwchem_config NWCHEM_MODULES="all" if [ `uname -m` == "x86_64" ]; then make 64_to_32 fi make > make.log &

I updated this to include the list of patches they've released. There is still an annoying bug when using OpenMPI: passing in any arguments causes NWChem to segfault. The problem seems to be OpenMPI's use of "strlen" on the passed arguments, and NWChem not appending the trailing null character. ...MPICH2 does not have this same problem. Either you can rename the input file to "nwchem.nw" and run it without arguments (what I'm currently doing), or switch to MPICH (also in the AUR). For the latter you just need to update config.sh with the proper MPICH parameters from its pkg-config. Hopefully I will be able to patch this soon.

Thank you for creating a PKGBUILD for this. NWChem can be one of the more tedious to set up. More so than GAMESS, at least. I'm hoping that Arch linux can become a great computational chemistry distro.