<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">Dear Mohammad <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">You could try using the cudas runtime
      libraries provided with the pgi compiler instead the ones of your
      system. Besure to have all the paths correctly set and that no
      other conflicting bynary or library appears on the path before the
      right ones. <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">For example I used these settings (
      copied from the module files provided with compiler) <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">export HOME_PGI='/opt/pgi' # this
      depends on where you installed the compiler though <br>
    </div>
    <div class="moz-cite-prefix">export
      CC=$HOME_PGI/linux86-64-llvm/2019/bin/pgcc</div>
    <div class="moz-cite-prefix">export
      FC=$HOME_PGI/linux86-64-llvm/2019/bin/pgfortran</div>
    <div class="moz-cite-prefix">export
      F90=$HOME_PGI/linux86-64-llvm/2019/bin/pgf90 <br>
    </div>
    <div class="moz-cite-prefix">export
      F77=$HOME_PGI/linux86-64-llvm/2019/bin/pgf77 <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">export
      PATH=$HOME_PGI/linux86-64-llvm/2019/bin:$PATH <br>
    </div>
    <div class="moz-cite-prefix">export
      LD_LIBRARY_PATH=$HOME_PGI/linux86-64-llvm/2019/lib:$LD_LIBRARY_PATH
      <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">export
      PATH=$HOME_PGI/linux86-64-llvm/2019/mpi/openmpi-3.1.3/bin:$PATH <br>
    </div>
    <div class="moz-cite-prefix">export
LD_LIBRARY_PATH=$HOME_PGI/linux86-64-llvm/2019/mpi/openmpi-3.1.3/lib:$LD_LIBRARY_PATH
      <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">then run pgaccelinfo to obtain the
      compute capilities of your card ( those that yoy have to pass in
      the configure script with the option -with-cuda-cc, it it the last
      line of the output e.g. in my case: <br>
    </div>
    <div class="moz-cite-prefix">PGI Default Target:           
      -ta=tesla:cc50</div>
    <div class="moz-cite-prefix">means that I have to use the option
      --with-cuda-cc=50 <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">then run the configure ( again in my
      case ... ) <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">configure
      --with-cuda=$HOME_PGI/pgi/linux86-64-llvm/2019/cuda/9.2
      --with-cuda-runtime=9.2 --with-cuda-cc=50 --enable-openmp
      --with-scalapack=no</div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">the 2019 community edition provides
      versions 10.1 10.0 and 9.2 of cuda runtime libraries so you can
      chose the one which fits with your card. <br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">hope that is works for you <br>
    </div>
    <div class="moz-cite-prefix">kind regards - Pietro Delugas<br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">On 10/27/19 3:08 PM, Mohammad Moaddeli
      wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CANfJyHTOvh_dkSJGZp7Lv_3A=vV54OeKub_bxbBM__BeEj_c2A@mail.gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      <div dir="ltr">
        <div>Dear all,</div>
        <div><br>
        </div>
        <div>I am trying to install q-e-gpu-qe-gpu-6.4.1a1. The
          following PATHs are added in /etc/bashrc:</div>
        <div><br>
        </div>
        <div>####  CUDA  ####<br>
          export PATH=/usr/local/cuda-10.1/bin:$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: <a class="moz-txt-link-abbreviated" href="mailto:pci@0000:04:00.0">pci@0000:04:00.0</a><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: <a class="moz-txt-link-abbreviated" href="mailto:pci@0000:12:00.0">pci@0000:12:00.0</a><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"
            moz-do-not-send="true">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"
            moz-do-not-send="true">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" moz-do-not-send="true">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>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <pre class="moz-quote-pre" wrap="">_______________________________________________
Quantum ESPRESSO is supported by MaX (<a class="moz-txt-link-abbreviated" href="http://www.max-centre.eu/quantum-espresso">www.max-centre.eu/quantum-espresso</a>)
users mailing list <a class="moz-txt-link-abbreviated" href="mailto:users@lists.quantum-espresso.org">users@lists.quantum-espresso.org</a>
<a class="moz-txt-link-freetext" href="https://lists.quantum-espresso.org/mailman/listinfo/users">https://lists.quantum-espresso.org/mailman/listinfo/users</a></pre>
    </blockquote>
    <p><br>
    </p>
  </body>
</html>