[QE-users] Installation verification
Mahmood Naderan
mahmood.nt at gmail.com
Tue Apr 30 20:22:47 CEST 2019
I think I did all things correctly. In order to override the default mpirun
(which is 2.0.0) I ran:
# export
PATH=/usr/lib64/qt-3.3/bin:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/share/apps/softwares/openmpi-4.0.1/bin
# export LD_LIBRARY_PATH=/share/apps/softwares/openmpi-4.0.1/lib
# mpirun --version
mpirun (Open MPI) 4.0.1
Report bugs to http://www.open-mpi.org/community/help/
Then I tried to build qe
# ./configure --prefix=/share/apps/softwares/q-e-qe-6.4/build
# make all -j6
# make install
...
\u2018XSpectra/src/xspectra.x\u2019 ->
\u2018/share/apps/softwares/q-e-qe-6.4/build/bin/xspectra.x\u2019
Quantum ESPRESSO binaries are installed in
/share/apps/softwares/q-e-qe-6.4/build/bin
# ldd build/bin/pw.x
linux-vdso.so.1 => (0x00007ffd51f58000)
libmpi_usempi.so.40 =>
/share/apps/softwares/openmpi-4.0.1/lib/libmpi_usempi.so.40
(0x00007f16552c8000)
libmpi_mpifh.so.40 =>
/share/apps/softwares/openmpi-4.0.1/lib/libmpi_mpifh.so.40
(0x00007f1655070000)
libmpi.so.40 => /share/apps/softwares/openmpi-4.0.1/lib/libmpi.so.40
(0x00007f1654d58000)
libgfortran.so.3 => /lib64/libgfortran.so.3 (0x00007f1654a10000)
libm.so.6 => /lib64/libm.so.6 (0x00007f1654708000)
libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f16544f0000)
libquadmath.so.0 => /lib64/libquadmath.so.0 (0x00007f16542b0000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f1654090000)
libc.so.6 => /lib64/libc.so.6 (0x00007f1653cc8000)
libopen-rte.so.40 =>
/share/apps/softwares/openmpi-4.0.1/lib/libopen-rte.so.40
(0x00007f1653a08000)
libopen-pal.so.40 =>
/share/apps/softwares/openmpi-4.0.1/lib/libopen-pal.so.40
(0x00007f16536f8000)
libdl.so.2 => /lib64/libdl.so.2 (0x00007f16534f0000)
libudev.so.1 => /lib64/libudev.so.1 (0x00007f16534d8000)
librt.so.1 => /lib64/librt.so.1 (0x00007f16532d0000)
libutil.so.1 => /lib64/libutil.so.1 (0x00007f16530c8000)
libz.so.1 => /lib64/libz.so.1 (0x00007f1652eb0000)
/lib64/ld-linux-x86-64.so.2 (0x00005652f0f3b000)
libcap.so.2 => /lib64/libcap.so.2 (0x00007f1652ca8000)
libdw.so.1 => /lib64/libdw.so.1 (0x00007f1652a60000)
libattr.so.1 => /lib64/libattr.so.1 (0x00007f1652858000)
libelf.so.1 => /lib64/libelf.so.1 (0x00007f1652640000)
liblzma.so.5 => /lib64/liblzma.so.5 (0x00007f1652418000)
libbz2.so.1 => /lib64/libbz2.so.1 (0x00007f1652208000)
So, it seems that qe-6.4 has been built with openmpi-4 correctly.
Back to my user account, I again set PATH and LD_LIBRARY_PATH in order to
override the default openmpi on the system.
As I tried my own file, I received the following error
$ which mpirun
/share/apps/softwares/openmpi-4.0.1/bin/mpirun
$ mpirun --version
mpirun (Open MPI) 4.0.1
Report bugs to http://www.open-mpi.org/community/help/
$ which pw.x
/share/apps/softwares/q-e-qe-6.4/build/bin/pw.x
[mahmood at rocks7 job]$ mpirun -np 2 pw.x mos64.scf.in
--------------------------------------------------------------------------
As of version 3.0.0, the "sm" BTL is no longer available in Open MPI.
Efficient, high-speed same-node shared memory communication support in
Open MPI is available in the "vader" BTL. To use the vader BTL, you
can re-run your job with:
mpirun --mca btl vader,self,... your_mpi_application
--------------------------------------------------------------------------
--------------------------------------------------------------------------
A requested component was not found, or was unable to be opened. This
means that this component is either not installed or is unable to be
used on your system (e.g., sometimes this means that shared libraries
that the component requires are unable to be found/loaded). Note that
Open MPI stopped checking at the first component that it did not find.
Host: rocks7.jupiterclusterscu.com
Framework: btl
Component: sm
--------------------------------------------------------------------------
--------------------------------------------------------------------------
It looks like MPI_INIT failed for some reason; your parallel process is
likely to abort. There are many reasons that a parallel process can
fail during MPI_INIT; some of which are due to configuration or environment
problems. This failure appears to be an internal failure; here's some
additional information (which may only be relevant to an Open MPI
developer):
mca_bml_base_open() failed
--> Returned "Not found" (-13) instead of "Success" (0)
--------------------------------------------------------------------------
[rocks7:08912] *** An error occurred in MPI_Init
[rocks7:08912] *** reported by process [1787822081,3255307776955514881]
[rocks7:08912] *** on a NULL communicator
[rocks7:08912] *** Unknown error
[rocks7:08912] *** MPI_ERRORS_ARE_FATAL (processes in this communicator
will now abort,
[rocks7:08912] *** and potentially your MPI job)
[rocks7.jupiterclusterscu.com:08903] 1 more process has sent help message
help-mpi-btl-sm.txt / btl sm is dead
[rocks7.jupiterclusterscu.com:08903] Set MCA parameter
"orte_base_help_aggregate" to 0 to see all help / error messages
[rocks7.jupiterclusterscu.com:08903] 1 more process has sent help message
help-mca-base.txt / find-available:not-valid
[rocks7.jupiterclusterscu.com:08903] 1 more process has sent help message
help-mpi-runtime.txt / mpi_init:startup:internal-failure
[rocks7.jupiterclusterscu.com:08903] 1 more process has sent help message
help-mpi-errors.txt / mpi_errors_are_fatal unknown handle
>From this line
As of version 3.0.0, the "sm" BTL is no longer available in Open MPI.
I think that QE is using a feature from OpenMPI which is not available in
Openmpi-4. Do you confirm that? Should I use OpenMPI-3?
My input file is
&CONTROL
calculation = "scf"
forc_conv_thr = 1.00000e-03
max_seconds = 1.72800e+05
prefix = 'mos2',
verbosity = 'high' ,
tstress = .true. ,
tprnfor = .true. ,
nstep = 100
outdir = '/home/mahmood/qe/' ,
pseudo_dir = '/share/apps/softwaers/q-e-qe-6.4.1/pseudo/' ,
/
&SYSTEM
a = 3.16100e+00
c = 1.22950e+01
degauss = 1.00000e-02
ecutrho = 440
ecutwfc = 48
ibrav = 4
nat = 6
nspin = 2
ntyp = 2
occupations = "smearing"
smearing = "gaussian"
starting_magnetization(1) = 2.00000e-01
starting_magnetization(2) = 0.00000e+00
/
&ELECTRONS
conv_thr = 1.00000e-06
electron_maxstep = 200
mixing_beta = 7.00000e-01
startingpot = "atomic"
startingwfc = "atomic+random"
/
&IONS
ion_dynamics = "bfgs"
/
&CELL
cell_dynamics = "bfgs"
press_conv_thr = 5.00000e-01
/
K_POINTS {automatic}
4 4 1 0 0 0
ATOMIC_SPECIES
Mo 95.94000 Mo.revpbe-spn-rrkjus_psl.0.3.0.UPF
S 32.06600 S.revpbe-n-rrkjus_psl.0.1.UPF
ATOMIC_POSITIONS {angstrom}
Mo -0.000002917 1.825008211 3.073729342
Mo 1.580497606 0.912506275 9.221229342
S -0.000005246 1.825013176 7.657345580
S 1.580494470 0.912510775 1.509845580
S 0.000001175 1.825005054 10.785175078
S 1.580498911 0.912501509 4.637675078
Regards,
Mahmood
On Tue, Apr 30, 2019 at 5:20 PM Paolo Giannozzi <p.giannozzi at gmail.com>
wrote:
> The MPI run-time environment (e.g. mpirun) must be consistent with the MPI
> libraries you are linking. Be very careful when setting your PATH, LIBPATH,
> etc. environment variables. They must be the same when you configure,
> compile, and run.
>
> Paolo
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190430/f5276d1f/attachment.html>
More information about the users
mailing list