Compiling error for 77_ddb/nmsq_pure_gkk_sumfs.F90 in SUSE

[vistawanted]

Dear all:
I am compiling Abinit 6.12.3 in OpenSUSE 12.1 x86_64 by icc+ifort+openmpi, where: the version of intel compilers is 12.1.5 (20120612) and the openmpi is 1.6.
The configure is as:

Code: Select all

./configure --prefix=/home/cmc/bin/Abinit FC=mpif90 CC=mpiCC CXX=mpiCC --enable-mpi=yes --enable-mpi-io=no --enable-vdwxc=yes --with-fft-flavor="fftw3" --with-fft-incs="-I${intel}/mkl/include/fftw" --with-fft-libs="-L${intel}/mkl/interfaces/fftw3xf/ -lfftw3xf_intel" --with-linalg-flavor="mkl" --with-linalg-incs="-I${intel}/mkl/include/intel64/lp64/ -I${intel}/mkl/include/" --with-linalg-libs="-L${intel}/mkl/lib/intel64/ -lmkl_intel_lp64 -lmkl_core -lmkl_sequential -lmkl_blas95_lp64 -lmkl_lapack95_lp64" --with-dft-flavor="libxc+wannier90"

Up to 77_ddb, I obtained such an error report:

Code: Select all

mpif90 -DHAVE_CONFIG_H -I. -I../.. -I../../src/incs -I../../src/incs -I${installdir}/abinit-6.12.3/fallbacks/exports/include    -free -module ${installdir}/abinit-6.12.3/src/mods -O2 -xHost -g -extend-source -vec-report0 -noaltparam -nofpscomp  -c -o thm9.o thm9.F90
���������������������������������������������������������������......���������������������������������������������������������������������������������������������������������������EEEEEE : catastrophic error: **Internal compiler error: segmentation violation signal raised** Please report this error along with the circumstances in which it occurred in a Software Problem Report.  Note: File and line given may not be explicit cause of this error.
compilation aborted for nmsq_pure_gkk_sumfs.F90 (code 1)
make[5]: *** [nmsq_pure_gkk_sumfs.o] Error 1
make[5]: *** Waiting for unfinished jobs....
make[5]: Leaving directory `/home/cmc/Downloads/usr/abinit-6.12.3/src/77_ddb'
make[4]: *** [all-recursive] Error 1

It is truely shown a long array of "squred ?".
As a result, I compiled this file seperately without the parameter "-xHost", then success...... But, I am afraid whether this "trick" introduces some unstable for my further using.....

[jbeuken]

Hi,

we have a bot with CentOS 6.3 / Xeon W3520 and ifort 12.1.4 with openmpi 1.6.0

we compile with the same options ( -O2 -xHost -g -extend-source -vec-report0 -noaltparam -nofpscomp ) and it works :

Code: Select all

[buildbot@coba2 77_ddb]$ ifort -V
Intel(R) Fortran Intel(R) 64 Compiler XE for applications running on Intel(R) 64, Version 12.1.4.319 Build 20120410
Copyright (C) 1985-2012 Intel Corporation.  All rights reserved.
FOR NON-COMMERCIAL USE ONLY

[buildbot@coba2 77_ddb]$ mpif90 -show
ifort -I/usr/local/openmpi-1.6.0-intel-12.1/include -I/usr/local/openmpi-1.6.0-intel-12.1/lib -shared-intel -L/usr/local/openmpi-1.6.0-intel-12.1/lib -lmpi_f90 -lmpi_f77 -lmpi -lrt -lnsl -lutil -ldl -lm -Wl,--export-dynamic -lrt -lnsl -lutil

[buildbot@coba2 77_ddb]$ rm thm9.o

[buildbot@coba2 77_ddb]$ make

/usr/local/openmpi_intel12/bin/mpif90 -DHAVE_CONFIG_H -I. -I../.. -I../../src/incs -I../../src/incs -I/home/buildbot/ABINIT/coba2_intel12/trunk_6.15.2-private/fallbacks/exports/include -I/home/buildbot/ABINIT/coba2_intel12/trunk_6.15.2-private/fallbacks/exports/include   -free -module /home/buildbot/ABINIT/coba2_intel12/trunk_6.15.2-private/src/mods -O2 -xHost -g -extend-source -vec-report0 -noaltparam -nofpscomp  -c -o thm9.o thm9.F90
rm -f lib77_ddb.a
ar rc lib77_ddb.a alignph.o anaddb_dtset_clean.o anaddb_dtset_nullify.o asrif9.o axial9.o bigbx9.o canat9.o canct9.o carteig2d.o carttransf.o chki8.o chkin9.o chkr8.o chkrp9.o cmpar8.o complete_gamma.o complete_gamma_tr.o complete_gkk.o completeperts.o d2c_weights.o d2c_wtq.o defs_elphon.o diel9.o dist9.o dtchi.o dtech9.o dymfz9.o elast9.o electrooptic.o eli_app_m_1d.o eli_diag_m_1d.o eli_lambda_1d.o eli_m_iter_1d.o eli_z_1d.o eliashberg_1d.o elphon.o ep_el_weights.o ep_fs_weights.o ep_ph_weights.o ep_setupqpt.o ewald9.o freeze_displ_allmodes.o ftgam.o ftgkk.o ftifc_q2r.o ftifc_r2q.o fxgkkphase.o gamma9.o gam_mult_displ.o get_all_gkk2.o get_all_gkq.o get_all_gkr.o get_fs_bands.o get_veloc_tr.o get_nv_fs_en.o get_nv_fs_temp.o gtblk9.o gtdyn9.o hybrid9.o ifclo9.o init8.o init9.o inpphon.o instr9.o integrate_gamma.o integrate_gamma_alt.o integrate_gamma_tr.o integrate_gamma_tr_lova.o interpolate_gkk.o invars9.o k_neighbors.o lin_interpq_gam.o mblktyp1.o mblktyp5.o mka2f.o mka2fQgrid.o mka2f_tr.o mka2f_tr_lova.o mkfskgrid.o mkfsqgrid.o mkifc9.o mkphbs.o m_atprj.o m_eph.o m_gamma.o m_ifc.o m_io_gkk.o m_phdos.o m_phonon_supercell.o m_primcell_ddb_info.o m_prtscphon.o m_sortph.o mkph_linwid.o mk_irredpert.o mkqptequiv.o nanal9.o new_integrate_gamma.o new_integrate_gamma_tr.o new_integrate_gamma_tr_lova.o nmsq_gam_sumfs.o nmsq_gam.o nmsq_pure_gkk.o nmsq_pure_gkk_sumfs.o normsq_gkq.o omega_decomp.o order_fs_kpts.o outelph.o outg2f.o outlwf9.o outphdos.o outvars9.o piezo9.o prt_gkk_yambo.o prtvsound.o ramansus.o rchkgsheader.o rdddb9.o read_el_veloc.o read_gkk.o refineblk.o relaxpol.o rsiaf9.o symdm9.o symgamma.o test_ftgkk.o thmeig.o thm9.o wght9.o interfaces_77_ddb.o 
ranlib lib77_ddb.a

bug in the 12.1.5 version of ifort ?

regards

jmb

[vistawanted]

Dear jbeuken:
Thank you for your kind reply.
I tried the same compiling in CentOS 5.6, where icc+ifort 12.1.5 is also installed with OpenMPI 1.6.0.
The exactly same configure parameter was also used.
But, the problem did not occur and all of the compiling process just worked smoothly.
I think such a compiling error may be rised from the incompatibility between OpenSUSE 12.1 and ifort 12.1.5.