Serial version of QE is compiled instead of parallel Running ./configure in the QE directory should automatically find all the necessary libraries needed for compiling the QE. It chooses the best version available on your computer or node etc., provided those libraries are in the standard path. In most of the cases in custom built clusters, packages are not maintained by the OS package management. It may have been compiled and may be present in the non-standard directories Even if the path to libraries are mentioned in LD_PATH, sometimes auto configure fails to recognize correct path to libraries installed on the computer. In the present situation ./congifure is failing to recognize the correct path to MPI libraries. You can add the MPI library path during configuration with the MPI_LIBS="/path/to/library/" or change the MPI_LIBS= "/path/to/library/" in make.sys after running the ./congifure . At the end of the...
Posts
Signal 7 (Bus error)
- Get link
- X
- Other Apps
Signal 7 (Bus error) Few days back I got a similar error, a bit of googling showed it may not be the problem of quantum espresso but may be of MPI. Contrary to that MPI forum said it could be the problem with the program but not with MPI. I somehow convinced myself that it may be due to the problem with quantum espresso because of newly installed node was giving this error. Hence, I recompiled quantum espresso on that node, to my surprise it solve the problem temporarily. Again after few days later I got the following output on my cluster. Taking the previous solution I recompiled the program, but for my surprise, I got the same error again. Digging into the cluster and checking the hard disk memory available, revealed that there was no memory left. It happens often because of not removing the memory intensive wave function files saved during previous calculations. Freeing the memory solved the problem and quantum espresso is ...
LAPACK compilation “error : no target found”
- Get link
- X
- Other Apps
LAPACK compilation “error : no target found” EDIT: Grab a new copy of pwscf and retry this error may not reappear. I recently encountered this on a system containing intel cpu ( it is not a problem of cpu, but of a compiler). I tried to compile with fresh package and found to be working fine. updated on 10/11/2015 During pwscf compilation, rarely we can encounter this kind of error. I have witnessed this error two times with Intel CPU and ubuntu 14.04 with nvidia graphic cards. This error may arise due to compiler problems, adding compiler option in the make.inc f ile solves the problem. make.inc file is present in espresso-5.1/lapack-3.2/ in the make.inc file you will see like the following # FORTRAN = gfortran OPTS = -O3 -g DRVOPTS = -O3 -g NOOPT = -O0 -g LOADER = mpif9...
Running pwscf/Quantum ESPRESSO on Beowulf like clusters
- Get link
- X
- Other Apps
Running pwscf/Quantum ESPRESSO on Beowulf like clusters This post may be helpful for beginners for using pwscf/ quantum espresso on Beowulf like Linux clusters. This post assumes that you have already built or have an access to it. In case you want to build it one, click here, but this need some changes and some tricks involved, which I will be posting next. To successfully run pwscf/ quantum espresso on cluster we need to follow new steps 1) Installing necessary libraries and openmpi 2) Installing pwscf/ quantum espresso 3) configuring ssh for mpi to access other nodes without password 4) Putting pseudo potential file on all nodes 5) executing the pwscf/ quantum espresso for calculation From here we will take one by one Step 1) Installing necessary libraries and openmpi I have used Ubuntu hence I will use apt-get for installing form terminal, if you have other distribution you can follow as recommended. sudo apt-get install openmpi-bin openmpi-common libopenmpi1.6 ...
Easy and efficient way of creating surface/slabs for pwscf/quantum espresso
- Get link
- X
- Other Apps
Easy and efficient way of creating surface/slabs for pwscf/quantum espresso There are a few ways of creating surfaces for pwscf/quantum espresso input files. It is easy to write coordinates for surface manually, provided you know crystallography. Writing unusual surfaces may be difficult. Hence there one easy way of creating surfaces for beginners. Atomic Simulation Environment (ASE) is a toolbox consisting of several tools and python modules can be used for setting up, manipulating, running, visualizing and analyzing atomistic simulations. First of all we need to set up ASE environment, the code is freely available under the GNU LGPL license . Install the ASE as mentioned in the ASE home page and run all the tests. To create CuZn alloy 211 surface I will use https://wiki.fysik.dtu.dk/ase/ase/surface.html?highlight=surface#ase.lattice.surface web-page as a reference and you can find more details for creating various surfaces there. And here I will only emphasis on ex...
- Get link
- X
- Other Apps
Q. How to create input file from cif files? Ans. First of all, why cif files to be used to create input file for QE. Because bigger unit cells with more than 3 types atoms are difficult to write by hand and to verify the input (at least for me). Cif files consists of everything related to crystallography of the material of the interest but it is not yet ready for the in put for QE. Lets take an example how to prepare input from cif file called EDI.cif (available from IZA website) Step 1. copy the EDI.cif to ~(path to espresso)/espresso-xx.x/PW/tools Step 2. type "./cif2qe.sh EDI > EDI.in" by typing this you will be invocking a script to convert cif files to QE input file (dont add cif extenction for EDI in put and give .in extenction for out put) Step 3. You should see EDI.in QE input file in the folder and open this file, you can observe that the scpipt had printed many option for you which is required for QE calculation but this is not complete an exhausti...
ACML ERROR with LAPACK
- Get link
- X
- Other Apps
Q. I have used ACML and compiled successfully but while running pwscf error *** ACML error : ------- 4 invalid parameter -----. Why do I get this error. Ans. Most probably you have AMD APU ( I had this and got same error). Old version of ACML probably may not be compatible with pwscf and AMDAPU combination. Updating the ACML 6.0.6 has solved the problem for me. ACML 6.0.6 or higher are support for heterogeneous computing. It used CL version of BLAS (clblas) and FFT (clfft) libraries which transfers your CPU load to GPU and you get enhanced performance. example code to compile pwscf with opencl is as follows: ./configure BLAS_LIBS="/opt/clAmdBlas1.10.321/lib64/libclAmdBlas.so.1.10.321" FFT_LIBS="/opt/clAmdFft-1.10.321/lib64/libclAmdFft.Runtime.so.1.10.321" LAPACK_LIBS="-L/opt/acml-6.06/gfortran64_mp/lib/x86_64 -lacml_mp" I have used latest version of clAmdBlas, and also clAmdFft and given path to shared library for LAPACK I have given path to...