[QE-users] [QE-GPU]QE-GPU 6.5a1 configuration failed

[Pietro Bonfa]

Dear Kefan,

here's my guess:

1. most definitely you don't need "sudo" to configure (nor to compile),

2. the configure script is detecting you system-wide installation of 
OpenMPI that wraps gfortran instead of pgf90 and, as a consequence, 
gfortran is used in all subsequent steps.

You have two options:

1. try a serial build, that makes sense since you have only one GPU on 
your system. The configure command should look like this:

./configure F90=pgf90 FC=pgf90 --disable-parallel --with-cuda=$CUDA_HOME 
--with-cuda-runtime=10.2 --with-cuda-cc=75 --enable-openmp

2. use openmpi provided by PGI, which is located here

/opt/pgi/linux86-64/19.10/mpi/openmpi-3.1.3/

(you need to add all relevant paths to PATH and LD_LIBRARY_PATH)
This will allow you to oversubscribe the GPU, but don't abuse this 
possibility.

If you still have problems, please attach the full output of the 
configure command.

Hope this helps,
best,
Pietro

On 2/11/20 1:32 AM, Kefan Chen wrote:
> Hi there,
> 
> I'm trying to compile QE-GPU 6.5a1 on my system but I failed with the 
> first step: run 'configure' executable. I followed the instructions 
> here:https://gitlab.com/QEF/q-e-gpu/-/wikis/home 
> <https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgitlab.com%2FQEF%2Fq-e-gpu%2F-%2Fwikis%2Fhome&data=02%7C01%7Cpietro.bonfa%40unipr.it%7C86e6c1fe10a14ebe49a708d7ae89f109%7Cbb064bc5b7a841ecbabed7beb3faeb1c%7C0%7C0%7C637169780270039551&sdata=wuQe9BjdEBuOrERCUOSLxAhDOAVTbQXxsEE7W9CjOjg%3D&reserved=0>
> 
> The error Info I have, the full info of running 'configure' can be seen 
> at the end of the email:
> 
> checking whether Fortran compiler accepts -Mcuda... no
> configure: error: You do not have the cudafor module. Are you using a 
> PGI compiler?
> 
> The command line I used to run 'configure':
> sudo ./configure --with-cuda=$CUDA_HOME --with-cuda-runtime=10.2 
> --with-cuda-cc=75 --enable-openmp
> 
> CUDA toolkit was installed previously and individually, which is 
> functional since I compiled LAMMPS with that. I have an RTX2080 Ti GPU, 
> whose architecture is Turing, so cuda-cc=75. I also tried to change 
> $CUDA_HOME to /usr/bin/cuda(where I installed cuda) but the error 
> persists.My OpenMPI was also installed separately, and the version is  4.0.2
> 
> I installed PGI 19.10 under /opt/pgi so I modified my 
> environment variable with following command lines.
> 
> export LD_LIBRARY_PATH=/opt/pgi/linux86-64/19.10/include:$LD_LIBRARY_PATH
> export PATH=/opt/pgi/linux86-64/19.1/bin:$PATH
> export LM_LICENSE_FILE=$LM_LICENSE_FILE:/opt/pgi/license.dat
> export MANPATH=$MANPATH:/opt/pgi/linux86-64/19.1/man
> 
> export CC=/opt/pgi/linux86-64/19.10/bin/pgcc
> export CXX=/opt/pgi/linux86-64/19.10/bin/pgc++
> export F90=/opt/pgi/linux86-64/19.10/bin/pgfortran
> export F77=/opt/pgi/linux86-64/19.10/bin/pgfortran
> export FC=/opt/pgi/linux86-64/19.10/bin/pgfortran
> export PGI=/opt/pgi
> 
> Do I miss any step? Thank you for your help!
> 
> BTW, I tried running configure only without any extra flags. It turned 
> out to be successful.
> 
> Regards,
> 
> Kefan Chen
> Ph.D student
> Department of Chemical & Material Engineering
> University of Nevada, Reno
> 
> 
> Full info of running 'configure':
> 
> checking build system type... x86_64-pc-linux-gnu
> checking ARCH... x86_64
> checking setting AR... ... ar
> checking setting ARFLAGS... ... ruv
> checking for gfortran... gfortran
> checking whether the Fortran compiler works... yes
> checking for Fortran compiler default output file name... a.out
> checking for suffix of executables...
> checking whether we are cross compiling... no
> checking for suffix of object files... o
> checking whether we are using the GNU Fortran compiler... yes
> checking whether gfortran accepts -g... yes
> checking for mpif90... mpif90
> checking whether we are using the GNU Fortran compiler... yes
> checking whether mpif90 accepts -g... yes
> checking version of mpif90... gfortran 7.4
> checking for Fortran flag to compile .f90 files... none
> setting F90... gfortran
> setting MPIF90... mpif90
> checking for cc... cc
> checking whether we are using the GNU C compiler... yes
> checking whether cc accepts -g... yes
> checking for cc option to accept ISO C89... none needed
> setting CC... cc
> setting CFLAGS... -O3
> checking for gfortran... gfortran
> checking whether we are using the GNU Fortran 77 compiler... yes
> checking whether gfortran accepts -g... yes
> setting F77... gfortran
> using F90... gfortran
> setting FFLAGS... -O3 -g -fopenmp
> setting F90FLAGS... $(FFLAGS) -x f95-cpp-input -fopenmp
> setting FFLAGS_NOOPT... -O0 -g
> setting CPP... cpp
> setting CPPFLAGS... -P -traditional -Uvector
> setting LD... mpif90
> setting LDFLAGS... -g -pthread -fopenmp
> checking for Fortran flag to compile .f90 files... (cached) none
> checking whether Fortran compiler accepts -Mcuda... no
> configure: error: You do not have the cudafor module. Are you using a 
> PGI compiler?
> 
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (https://eur01.safelinks.protection.outlook.com/?url=www.max-centre.eu%2Fquantum-espresso&data=02%7C01%7Cpietro.bonfa%40unipr.it%7C86e6c1fe10a14ebe49a708d7ae89f109%7Cbb064bc5b7a841ecbabed7beb3faeb1c%7C0%7C0%7C637169780270059536&sdata=MNHqPeFTS4acVzLmfaPq2U2RI92xNFODA4ITc9DKZfU%3D&reserved=0)
> users mailing list users at lists.quantum-espresso.org
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusers&data=02%7C01%7Cpietro.bonfa%40unipr.it%7C86e6c1fe10a14ebe49a708d7ae89f109%7Cbb064bc5b7a841ecbabed7beb3faeb1c%7C0%7C0%7C637169780270069533&sdata=0%2FaPlzMksKsSEZGZWdRU7HS6ZYH4cyQeZ8Ly9Qw0cFE%3D&reserved=0
>