<div dir="ltr"><div>Dear all,</div><div><br></div><div>I am trying to install q-e-gpu-qe-gpu-6.4.1a1. The following PATHs are added in /etc/bashrc:</div><div><br></div><div>####  CUDA  ####<br>export PATH=/usr/local/cuda-10.1/bin:$PATH<br>export PATH=/usr/local/cuda-10.1/include:$PATH<br>export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64:$LD_LIBRARY_PATH<br>export LD_LIBRARY_PATH=/usr/local/cuda-10.1/extras/CUPTI/lib64:$LD_LIBRARY_PATH</div><div><br>####  PGI  ####<br>PGI=/opt/pgi<br>export PGI<br>PATH=/opt/pgi/linux86-64/19.4/bin:$PATH<br>export PATH<br>PATH=/opt/pgi/linux86-64/19.4/mpi/openmpi-3.1.3/bin:$PATH<br>export PATH<br>PATH=/opt/pgi/linux86-64/19.4/mpi/openmpi-3.1.3/include:$PATH<br>export PATH<br>PATH=/opt/pgi/linux86-64/19.4/mpi/openmpi-3.1.3/lib:$PATH<br>export PATH<br>MANPATH=$MANPATH:/opt/pgi/linux86-64/19.4/man<br>export MANPATH</div><div><br></div><div>and also the Graphic Card driver is installed:<br><br>[moaddeli@localhost ~]$ nvidia-smi<br>Sun Oct 27 16:45:25 2019<br>+-----------------------------------------------------------------------------+<br>| NVIDIA-SMI 430.50       Driver Version: 430.50       CUDA Version: 10.1     |<br>|-------------------------------+----------------------+----------------------+<br>| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |<br>| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |<br>|===============================+======================+======================|<br>|   0  GeForce GTX 108...  Off  | 00000000:04:00.0  On |                  N/A |<br>| 50%   28C    P8    10W / 250W |     58MiB / 11175MiB |      0%      Default |<br>+-------------------------------+----------------------+----------------------+<br><br>+-----------------------------------------------------------------------------+<br>| Processes:                                                       GPU Memory |<br>|  GPU       PID   Type   Process name                             Usage      |<br>|=============================================================================|<br>|    0     14528      G   /usr/bin/X                                    39MiB |<br>|    0     14584      G   /usr/bin/gnome-shell                          16MiB |<br>+-----------------------------------------------------------------------------+<br><br><br>[root@localhost moaddeli]# lshw -numeric -C display<br>  *-display<br>       description: VGA compatible controller<br>       product: GP102 [GeForce GTX 1080 Ti] [10DE:1B06]<br>       vendor: NVIDIA Corporation [10DE]<br>       physical id: 0<br>       bus info: pci@0000:04:00.0<br>       version: a1<br>       width: 64 bits<br>       clock: 33MHz<br>       capabilities: pm msi pciexpress vga_controller bus_master cap_list rom<br>       configuration: driver=nvidia latency=0<br>       resources: iomemory:27f0-27ef iomemory:27f0-27ef irq:89 memory:c4000000-c4ffffff memory:27fe0000000-27fefffffff memory:27ff0000000-27ff1ffffff ioport:c000(size=128) memory:c5000000-c507ffff<br>  *-display<br>       description: VGA compatible controller<br>       product: ASPEED Graphics Family [1A03:2000]<br>       vendor: ASPEED Technology, Inc. [1A03]<br>       physical id: 0<br>       bus info: pci@0000:12:00.0<br>       version: 30<br>       width: 32 bits<br>       clock: 33MHz<br>       capabilities: pm msi vga_controller cap_list<br>       configuration: driver=ast latency=0<br>       resources: irq:16 memory:c6000000-c6ffffff memory:c7000000-c701ffff ioport:b000(size=128)<br><br></div><div>When I compile the code in a serial version, the executable pw.x is created in bin directory and seems to work well:</div><div><br></div><div>
[moaddeli@localhost test]$ /codes/qe4/q-e-gpu-qe-gpu-6.4.1a1/bin/pw.x <br></div><div><br></div><div>     Program PWSCF v.6.4.1 starts on 27Oct2019 at 16:48: 2<br><br>     This program is part of the open-source Quantum ESPRESSO suite<br>     for quantum simulation of materials; please cite<br>         "P. Giannozzi et al., J. Phys.:Condens. Matter 21 395502 (2009);<br>         "P. Giannozzi et al., J. Phys.:Condens. Matter 29 465901 (2017);<br>          URL <a href="http://www.quantum-espresso.org">http://www.quantum-espresso.org</a>",<br>     in publications or presentations arising from this work. More details at<br>     <a href="http://www.quantum-espresso.org/quote">http://www.quantum-espresso.org/quote</a><br><br>     Serial version<br>     Waiting for input...<br><br></div><div>however, the following error appears:<br><br>[moaddeli@localhost test]$ /codes/qe4/q-e-gpu-qe-gpu-6.4.1a1/bin/pw.x <<a href="http://c.in">c.in</a> | tee c.out<br>0: ALLOCATE: copyin Symbol Memcpy FAILED:13(invalid device symbol)<br><br><br> 
When I compile the code in a parallel version,


the executable pw.x is created in bin directory, but it does not work:<br><br>[moaddeli@localhost ~]$ /codes/qe4/q-e-gpu-qe-gpu-6.4.1a1/bin/pw.x<br>[localhost.localdomain:40889] [[INVALID],INVALID] ORTE_ERROR_LOG: A system-required executable either could not be found or was not executable by this user in file ../../../../../orte/mca/ess/singleton/ess_singleton_module.c at line 388<br>[localhost.localdomain:40889] [[INVALID],INVALID] ORTE_ERROR_LOG: A system-required executable either could not be found or was not executable by this user in file ../../../../../orte/mca/ess/singleton/ess_singleton_module.c at line 166<br>--------------------------------------------------------------------------<br>Sorry!  You were supposed to get help about:<br>    orte_init:startup:internal-failure<br>But I couldn't open the help file:<br>    /proj/pgi/linux86-64-llvm/2019/mpi/openmpi-3.1.3/share/openmpi/help-orte-runtime: No such file or directory.  Sorry!<br>--------------------------------------------------------------------------<br>--------------------------------------------------------------------------<br>Sorry!  You were supposed to get help about:<br>    mpi_init:startup:internal-failure<br>But I couldn't open the help file:<br>    /proj/pgi/linux86-64-llvm/2019/mpi/openmpi-3.1.3/share/openmpi/help-mpi-runtime.txt: No such file or directory.  Sorry!<br>--------------------------------------------------------------------------<br>*** An error occurred in MPI_Init<br>*** on a NULL communicator<br>*** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,<br>***    and potentially your MPI job)<br>[localhost.localdomain:40889] Local abort before MPI_INIT completed completed successfully, but am not able to aggregate error messages, and not able to guarantee that all other processes were killed!<br></div><div><br></div><div>Any help will be greatly appreciated.</div><div><br></div><div>Mohammad Moaddeli</div><div><br></div><div>Shiraz University</div><div><br></div></div>