<div dir="ltr"><div>Intel DFTI FFT's are contained in the Intel MKL libraries. You should link them if you have them installed<br><br></div>Paolo<br></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Oct 27, 2016 at 8:10 PM, Downs, Andrew S CTR USARMY ARL (US) <span dir="ltr"><<a href="mailto:andrew.s.downs3.ctr@mail.mil" target="_blank">andrew.s.downs3.ctr@mail.mil</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hello again,<br>
<br>
I did eventually get QE6.0 to build on our Cray XC-40 by setting the arch type to crayxt, and some other small changes, however, I can only get it to work with GCC. I'd really like to compile it with Intel (we have Intel 16) but for whatever reason, it seems like no matter what I change, I always get hung up on the same 'open socket' spot.<br>
<br>
Build output:<br>
<br>
ftn -openmp -o pw.x \<br>
pwscf.o libpw.a ../../Modules/libqemod.a ../../FFTXlib/libqefft.a ../../LAXlib/libqela.a /p/home/adowns/useraide/qe6/<wbr>qe-6.0//clib/clib.a /p/home/adowns/useraide/qe6/<wbr>qe-6.0//iotk/src/libiotk.a<br>
/p/home/adowns/useraide/qe6/<wbr>qe-6.0//clib/clib.a(sockets.o)<wbr>: In function `open_socket':<br>
sockets.c:(.text+0x88): warning: Using 'getaddrinfo' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking<br>
../../FFTXlib/libqefft.a(fft_<wbr>scalar.o): In function `fft_scalar_mp_cft_1z_':<br>
fft_scalar.f90:(.text+0x38f): undefined reference to `dfti_compute_backward_zz'<br>
fft_scalar.f90:(.text+0x74f): undefined reference to `dfti_compute_forward_zz'<br>
../../FFTXlib/libqefft.a(fft_<wbr>scalar.o): In function `fft_scalarcft_1z_mp_init_<wbr>dfti_':<br>
fft_scalar.f90:(.text+0x928): undefined reference to `dfti_create_descriptor_1d'<br>
fft_scalar.f90:(.text+0x953): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x97e): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x9aa): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x9d5): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0xa1c): undefined reference to `dfti_set_value_dblval'<br>
fft_scalar.f90:(.text+0xa48): undefined reference to `dfti_set_value_dblval'<br>
fft_scalar.f90:(.text+0xa6a): undefined reference to `dfti_commit_descriptor_<wbr>external'<br>
fft_scalar.f90:(.text+0xc16): undefined reference to `dfti_free_descriptor_<wbr>external'<br>
../../FFTXlib/libqefft.a(fft_<wbr>scalar.o): In function `fft_scalar_mp_cft_2xy_':<br>
fft_scalar.f90:(.text+0xf1c): undefined reference to `dfti_compute_backward_z'<br>
fft_scalar.f90:(.text+0x1220): undefined reference to `dfti_compute_forward_z'<br>
../../FFTXlib/libqefft.a(fft_<wbr>scalar.o): In function `fft_scalarcft_2xy_mp_init_<wbr>dfti_':<br>
fft_scalar.f90:(.text+0x1474): undefined reference to `dfti_create_descriptor_highd'<br>
fft_scalar.f90:(.text+0x14a4): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x14ea): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x151b): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x1573): undefined reference to `dfti_set_value_dblval'<br>
fft_scalar.f90:(.text+0x15a4): undefined reference to `dfti_set_value_dblval'<br>
fft_scalar.f90:(.text+0x15cb): undefined reference to `dfti_commit_descriptor_<wbr>external'<br>
fft_scalar.f90:(.text+0x19d9): undefined reference to `dfti_free_descriptor_<wbr>external'<br>
../../FFTXlib/libqefft.a(fft_<wbr>scalar.o): In function `fft_scalar_mp_cfft3d_':<br>
fft_scalar.f90:(.text+0x1ebb): undefined reference to `dfti_compute_backward_z'<br>
fft_scalar.f90:(.text+0x2262): undefined reference to `dfti_compute_forward_z'<br>
../../FFTXlib/libqefft.a(fft_<wbr>scalar.o): In function `fft_scalarcfft3d_mp_init_<wbr>dfti_':<br>
fft_scalar.f90:(.text+0x2522): undefined reference to `dfti_create_descriptor_highd'<br>
fft_scalar.f90:(.text+0x2552): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x25a0): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x25d1): undefined reference to `dfti_set_value_intval'<br>
fft_scalar.f90:(.text+0x2637): undefined reference to `dfti_set_value_dblval'<br>
fft_scalar.f90:(.text+0x2674): undefined reference to `dfti_set_value_dblval'<br>
fft_scalar.f90:(.text+0x269b): undefined reference to `dfti_commit_descriptor_<wbr>external'<br>
fft_scalar.f90:(.text+0x2aad): undefined reference to `dfti_free_descriptor_<wbr>external'<br>
make[2]: *** [pw.x] Error 1<br>
make[2]: Leaving directory `/p/home/adowns/useraide/qe6/<wbr>qe-6.0/PW/src'<br>
make[1]: *** [pw] Error 1<br>
make[1]: Leaving directory `/p/home/adowns/useraide/qe6/<wbr>qe-6.0/PW'<br>
make: *** [pw] Error 1<br>
<br>
<br>
my configure line:<br>
./configure ARCH=crayxt MPIF90=ftn F77=ftn F90=ftn FFLAGS="-O3" CC=cc CFLAGS=-O3 --prefix=/usr/cta/unsupported/<wbr>qe/6.0_intel -enable-parallel -enable-openmp<br>
<br>
<br>
my make.inc file is as follows:<br>
<br>
# make.inc. Generated from <a href="http://make.inc.in" rel="noreferrer" target="_blank">make.inc.in</a> by configure.<br>
<br>
# compilation rules<br>
<br>
.SUFFIXES :<br>
.SUFFIXES : .o .c .f .f90<br>
<br>
# most fortran compilers can directly preprocess c-like directives: use<br>
# $(MPIF90) $(F90FLAGS) -c $<<br>
# if explicit preprocessing by the C preprocessor is needed, use:<br>
# $(CPP) $(CPPFLAGS) $< -o $*.F90<br>
# $(MPIF90) $(F90FLAGS) -c $*.F90 -o $*.o<br>
# remember the tabulator in the first column !!!<br>
<br>
.f90.o:<br>
$(MPIF90) $(F90FLAGS) -c $<<br>
<br>
# .f.o and .c.o: do not modify<br>
<br>
.f.o:<br>
$(F77) $(FFLAGS) -c $<<br>
<br>
.c.o:<br>
$(CC) $(CFLAGS) -c $<<br>
<br>
# Top QE directory, useful for locating libraries, linking QE with plugins<br>
# The following syntax should always point to TOPDIR:<br>
TOPDIR = $(dir $(abspath $(filter %make.inc,$(MAKEFILE_LIST))))<br>
# if it doesn't work, uncomment the following line (edit if needed):<br>
<br>
# TOPDIR = /p/home/adowns/useraide/qe6/<wbr>qe-6.0<br>
<br>
# DFLAGS = precompilation options (possible arguments to -D and -U)<br>
# used by the C compiler and preprocessor<br>
# FDFLAGS = as DFLAGS, for the f90 compiler<br>
# See include/defs.h.README for a list of options and their meaning<br>
# With the exception of IBM xlf, FDFLAGS = $(DFLAGS)<br>
# For IBM xlf, FDFLAGS is the same as DFLAGS with separating commas<br>
<br>
# MANUAL_DFLAGS = additional precompilation option(s), if desired<br>
# BEWARE: it does not work for IBM xlf! Manually edit FDFLAGS<br>
MANUAL_DFLAGS =<br>
DFLAGS = -D__OPENMP -D__INTEL -D__DFTI -D__MPI -D__SCALAPACK<br>
FDFLAGS = $(DFLAGS) $(MANUAL_DFLAGS)<br>
<br>
# IFLAGS = how to locate directories with *.h or *.f90 file to be included<br>
# typically -I../include -I/some/other/directory/<br>
# the latter contains .e.g. files needed by FFT libraries<br>
<br>
IFLAGS = -I$(TOPDIR)/include -I../include/ -I/opt/intel/compilers_and_<wbr>libraries_2016.0.109/linux/<wbr>mkl/include<br>
<br>
# MOD_FLAGS = flag used by f90 compiler to locate modules<br>
# Each Makefile defines the list of needed modules in MODFLAGS<br>
<br>
MOD_FLAG = -I<br>
<br>
# Compilers: fortran-90, fortran-77, C<br>
# If a parallel compilation is desired, MPIF90 should be a fortran-90<br>
# compiler that produces executables for parallel execution using MPI<br>
# (such as for instance mpif90, mpf90, mpxlf90,...);<br>
# otherwise, an ordinary fortran-90 compiler (f90, g95, xlf90, ifort,...)<br>
# If you have a parallel machine but no suitable candidate for MPIF90,<br>
# try to specify the directory containing "mpif.h" in IFLAGS<br>
# and to specify the location of MPI libraries in MPI_LIBS<br>
<br>
MPIF90 = ftn<br>
#F90 = ftn<br>
CC = cc<br>
F77 = ftn<br>
<br>
# C preprocessor and preprocessing flags - for explicit preprocessing,<br>
# if needed (see the compilation rules above)<br>
# preprocessing flags must include DFLAGS and IFLAGS<br>
<br>
CPP = cpp<br>
CPPFLAGS = -P -traditional $(DFLAGS) $(IFLAGS)<br>
<br>
# compiler flags: C, F90, F77<br>
# C flags must include DFLAGS and IFLAGS<br>
# F90 flags must include MODFLAGS, IFLAGS, and FDFLAGS with appropriate syntax<br>
<br>
CFLAGS = -O3 $(DFLAGS) $(IFLAGS)<br>
F90FLAGS = $(FFLAGS) -nomodule -openmp -fpp $(FDFLAGS) $(IFLAGS) $(MODFLAGS)<br>
FFLAGS = -O3<br>
<br>
# compiler flags without optimization for fortran-77<br>
# the latter is NEEDED to properly compile dlamch.f, used by lapack<br>
<br>
FFLAGS_NOOPT = -O0 -assume byterecl -g -traceback<br>
<br>
# compiler flag needed by some compilers when the main program is not fortran<br>
# Currently used for Yambo<br>
<br>
FFLAGS_NOMAIN = -nofor_main<br>
<br>
# Linker, linker-specific flags (if any)<br>
# Typically LD coincides with F90 or MPIF90, LD_LIBS is empty<br>
<br>
LD = ftn<br>
LDFLAGS = -openmp<br>
LD_LIBS =<br>
<br>
# External Libraries (if any) : blas, lapack, fft, MPI<br>
<br>
# If you have nothing better, use the local copy :<br>
# BLAS_LIBS = /your/path/to/espresso/BLAS/<wbr>blas.a<br>
# BLAS_LIBS_SWITCH = internal<br>
<br>
BLAS_LIBS =<br>
BLAS_LIBS_SWITCH = external<br>
<br>
# If you have nothing better, use the local copy :<br>
# LAPACK_LIBS = /your/path/to/espresso/lapack-<wbr>3.2/lapack.a<br>
# LAPACK_LIBS_SWITCH = internal<br>
# For IBM machines with essl (-D__ESSL): load essl BEFORE lapack !<br>
# remember that LAPACK_LIBS precedes BLAS_LIBS in loading order<br>
<br>
LAPACK_LIBS =<br>
LAPACK_LIBS_SWITCH = external<br>
<br>
SCALAPACK_LIBS =<br>
<br>
# nothing needed here if the the internal copy of FFTW is compiled<br>
# (needs -D__FFTW in DFLAGS)<br>
<br>
FFT_LIBS =<br>
<br>
# HDF5 -- experimental<br>
HDF5_LIB =<br>
<br>
# For parallel execution, the correct path to MPI libraries must<br>
# be specified in MPI_LIBS (except for IBM if you use mpxlf)<br>
<br>
MPI_LIBS =<br>
<br>
# IBM-specific: MASS libraries, if available and if -D__MASS is defined in FDFLAGS<br>
<br>
MASS_LIBS =<br>
<br>
# ar command and flags - for most architectures: AR = ar, ARFLAGS = ruv<br>
<br>
AR = ar<br>
ARFLAGS = ruv<br>
<br>
# ranlib command. If ranlib is not needed (it isn't in most cases) use<br>
# RANLIB = echo<br>
<br>
RANLIB = ranlib<br>
<br>
# all internal and external libraries - do not modify<br>
<br>
FLIB_TARGETS = all<br>
<br>
LIBOBJS = $(TOPDIR)/clib/clib.a $(TOPDIR)/iotk/src/libiotk.a<br>
LIBS = $(SCALAPACK_LIBS) $(LAPACK_LIBS) $(FFT_LIBS) $(BLAS_LIBS) $(MPI_LIBS) $(MASS_LIBS) ${HDF5_LIB} $(LD_LIBS)<br>
<br>
# wget or curl - useful to download from network<br>
WGET = wget -O<br>
<br>
# Install directory - not currently used<br>
PREFIX = /usr/cta/unsupported/qe/6.0_<wbr>intel<br>
<br>
<br>
______________________________<wbr>_________________<br>
Pw_forum mailing list<br>
<a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>
<a href="http://pwscf.org/mailman/listinfo/pw_forum" rel="noreferrer" target="_blank">http://pwscf.org/mailman/<wbr>listinfo/pw_forum</a><br>
</blockquote></div><br><br clear="all"><br>-- <br><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div>Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,<br>Univ. Udine, via delle Scienze 208, 33100 Udine, Italy<br>Phone +39-0432-558216, fax +39-0432-558222<br><br></div></div></div></div></div>
</div>