Can you be more specific? Is it your cluster, it is your compilation of Quantum Espresso? Not all cluster are created equal so...

Cheers,

Roberto

Actuallu due to lack of computational resources at our campus we are trying to use the cluster from other university  whose acess is provided to us. But facing problem to run Quantum espresso in parallel with given number of cores

I am afraid if you wish any help you really need to provide a few pieces of information. For example, do you have the output of Quantum Espresso? The input? The script you have used to run QE on your cluster? The error or output produced by the cluster.

In general, if QE is compiled to work on parallel machines, you have to instruct the cluster to run it. Common commands are srun and mpirun but it depends on the cluster software.

As it stands now, your question is too generic admittedly.

1- compile QE considering BLACS, ScaLAPACK, FFTW and ... and put it in a directory to be accessible from all nodes

2- create a file (e.g., hosts.txt) containing node names and the number of their cores per line:

node1:12

node2:12

node3:12

....

3- run QE based on your MPI version:

nohup mpirun -hosts ~/hosts.txt -n 36 pw.x < scf.in > scf.out &

or

4- make a job file based on the queuing system used in the cluster:

%cat test.pbs

#PBS -N test_qe

#PBS -q regular

#PBS –l nodes=3:ppn=12

#PBS –l walltime=72:00:00

#PBS -j oe

#PBS -V

cd $PBS_O_WORKDIR

module load qe

mpirun -n 36 pw.x < scf.in > scf.out

%qsub test.pbs

*I run a CP_QE job with a job submission script:

#!/bin/bash

#PBS -q S

#PBS -l select=1:ncpus=2:mpiprocs=2

#PBS -N qe_h2onacl

#PBS -o file.out

#PBS -j oe

#PBS -m b

source /etc/profile.d/modules.sh

module purge

module load intel

module load intelmpi

cd ${PBS_O_WORKDIR}

mpirun dplace ./cp.x -i h2onacl.in

*But it seems being not a parallel job due to output:

Program CP v.6.0 starts on 30Oct2018 at 13:54:46

This program is part of the open-source Quantum ESPRESSO suite

for quantum simulation of materials; please cite

"P. Giannozzi et al., J. Phys.:Condens. Matter 21 395502 (2009);

URL http://www.quantum-espresso.org",

in publications or presentations arising from this work. More details at

http://www.quantum-espresso.org/quote

Serial version

Dumping input parameters

Program CP v.6.0 starts on 30Oct2018 at 13:54:46

This program is part of the open-source Quantum ESPRESSO suite

for quantum simulation of materials; please cite

"P. Giannozzi et al., J. Phys.:Condens. Matter 21 395502 (2009);

URL http://www.quantum-espresso.org",

in publications or presentations arising from this work. More details at

http://www.quantum-espresso.org/quote

Serial version

Dumping input parameters

* How to repair this error ?

Dear Le Vang Sang,

I would say that your compilation of QE is a serial version, so the mistake happened at the compilation level. Who compiled QE for your cluster?

You can try

ldd cp.x

and check that it has been compiled with support for mpi?

Cheers,

Roberto

Dear Roberto,

Thank you for your kind comments. I submitted the job on a cluster of my university.

Yes. I will try to do as your comments.

Sincerely,

Sang

Dear Roberto,

* When running a CP job by Quantum Epresso on supercomputer with 80 cores, I

saw a result as

Program CP v.6.0 (svn rev. 13079) starts on 2Nov2018 at 8:32:19

This program is part of the open-source Quantum ESPRESSO suite

for quantum simulation of materials; please cite

"P. Giannozzi et al., J. Phys.:Condens. Matter 21 395502 (2009);

URL http://www.quantum-espresso.org",

in publications or presentations arising from this work. More details at

http://www.quantum-espresso.org/quote

Parallel version (MPI), running on 80 processors

R & G space division: proc/nbgrp/npool/nimage = 80

Dumping input parameters

......

* I think this job was run by a parallel process, but it was as slow as that run on CPU. Is it sure that the job is run parallel on 80 cores ? I set up a Qe-6.0 version.

Sincerely,

Sang

Dear Sang,

Yes, you see it from the lines

Parallel version (MPI), running on 80 processors

R & G space division: proc/nbgrp/npool/nimage = 80

This means that the job is running on 80 processors in parallel (I don’t know the structure of your cluster so I am not sure on how many nodes it is actually running). On terms of speed, again I cannot comment since I don’t know your cluster and the CPU it is made of.

Compare those two lines with the old case in which QE replied with the “Serial version’ to see the difference..

Kind regards,

Roberto

Dear Roberto,

Yes, thank you very much for your comments !

Sincerely,

Sang

Dear All,

I see a problem when setting Quantun-Espresso, it is an error at "fftw_interfaces.f90" (attacted below).

Please tell me how to repair it. Thanks !

...................................... attached........................................

The following libraries have been found:

BLAS_LIBS= -lmkl_gf_lp64 -lmkl_sequential -lmkl_core

LAPACK_LIBS=

FFT_LIBS=

Please check if this is what you expect.

If any libraries are missing, you may specify a list of directories

to search and retry, as follows:

./configure LIBDIRS="list of directories, separated by spaces"

Parallel environment detected successfully.\

Configured for compilation of parallel executables.

For more info, read the ESPRESSO User's Guide (Doc/users-guide.tex).

configure: success

levansang82@rokko1:/data/levansang82/q-e-qe-6.2.0> make all

test -d bin || mkdir bin

( cd FFTXlib ; make TLDEPS= all || exit 1 )

make[1]: Entering directory `/data/levansang82/q-e-qe-6.2.0/FFTXlib'

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c fft_param.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c fft_support.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c stick_base.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c fft_types.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c scatter_mod.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c fft_scalar.ARM_LIB.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c fft_scalar.DFTI.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c fft_scalar.ESSL.f90

mpif90 -O3 -g -x f95-cpp-input -D__DFTI -D__MPI -I/data/levansang82/q-e-qe-6.2.0//include -I/data/levansang82/q-e-qe-6.2.0//FoX/finclude -I../include/ -I/apps/intel/compilers_and_libraries_2018.2.199/linux/mkl/include -c fftw_interfaces.f90

fftw_interfaces.f90:34.37:

SUBROUTINE fft_x_stick( plan, xy, nx, ny, nz, ldx, ldy ) &

1

Error: C kind parameter is for type REAL but symbol 'xy' at (1) is of type COMPLEX

fftw_interfaces.f90:42.44:

SUBROUTINE fft_x_stick_single( plan, xy, nx, ny, nz, ldx, ldy ) &

1

Error: C kind parameter is for type REAL but symbol 'xy' at (1) is of type COMPLEX

fftw_interfaces.f90:50.37:

SUBROUTINE fft_y_stick( plan, xy, ny, ldx ) BIND(C,name="fft_y_stick")

1

Error: C kind parameter is for type REAL but symbol 'xy' at (1) is of type COMPLEX

fftw_interfaces.f90:57.36:

SUBROUTINE fft_z_stick( plan, z, ldz, nzl ) BIND(C,name="fft_z_stick")

1

Error: C kind parameter is for type REAL but symbol 'z' at (1) is of type COMPLEX

fftw_interfaces.f90:64.43:

SUBROUTINE fft_z_stick_single( plan, z, ldz ) BIND(C,name="fft_z_stick_sin

1

Error: C kind parameter is for type REAL but symbol 'z' at (1) is of type COMPLEX

fftw_interfaces.f90:71.47:

SUBROUTINE fftw_inplace_drv_1d( plan, m, z, inc1, inc2 ) &

1

Error: C kind parameter is for type REAL but symbol 'z' at (1) is of type COMPLEX

fftw_interfaces.f90:78.47:

SUBROUTINE fftw_inplace_drv_3d( plan, m, z, inc1, inc2 ) &

1

Error: C kind parameter is for type REAL but symbol 'z' at (1) is of type COMPLEX

make[1]: *** [fftw_interfaces.o] Error 1

make[1]: Leaving directory `/data/levansang82/q-e-qe-6.2.0/FFTXlib'

make: *** [libfft] Error 1

Ahmad Abdolmaleki I have created a jobscript in a similiar manner as you have mentioned . However i am receiving this error in the output file.

Hello Abdolmaleki,

The error in my installation is due to that I do not understand my suppercomputer.

After understanding it, the installation is OK!

The guides for installing in QE are OK, we should follow those step by step during installing. And, also understand our supercomputer.