Version 7.1.14 test seg fault

[S. David]

Hello,

I've tried to install Version 7.1.14 on my university's cluster. In the tests folder, mpi-test runs without issue, but the original test.sh fails one test (error message below). I wouldn't think this an issue since it's an mpi install, but I was checking it because I got the same error when running some of my old elk.in files (though not all). I re-used the make.inc recommended by cluster support from my last compiled version (4.3.6), with the addition of the OBLAS/BLIS recommendation from the forum.

Running test in directory test_022...
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
elk 0000000003FD3054 Unknown Unknown Unknown
libpthread-2.17.s 00002AADF7AD75E0 Unknown Unknown Unknown
libiomp5.so 00002AADF9568776 Unknown Unknown Unknown
libiomp5.so 00002AADF9568750 Unknown Unknown Unknown
libiomp5.so 00002AADF956ACCB Unknown Unknown Unknown
libiomp5.so 00002AADF954CB94 Unknown Unknown Unknown
ld-2.17.so 00002AADF78B3B58 Unknown Unknown Unknown
libc-2.17.so 00002AADF989FA69 Unknown Unknown Unknown
libc-2.17.so 00002AADF989FAB5 Unknown Unknown Unknown
libc-2.17.so 00002AADF9888C0C __libc_start_main Unknown Unknown
elk 0000000000406829 Unknown Unknown Unknown
Passed

MKLROOT = $(CURC_MKL_ROOT)
FFTROOT = $(CURC_FFTW_ROOT)
MAKE = make
F90 = mpiifort
F90_OPTS = -O3 -ip -xcore-avx2 -qopenmp -I$(MKLROOT)/include
F77 = mpiifort
F77_OPTS = -O3 -ip -xcore-avx2 -qopenmp
AR = ar
LIB_SYS =
LIB_LPK = -Wl,--start-group $(MKLROOT)/lib/intel64/libmkl_intel_lp64.a $(MKLROOT)/lib/intel64/libmkl_core.a $(MKLROOT)/lib/intel64/libmkl_intel_thread.a -Wl,--end-group -lpthread -lm -qopenmp $(FFTROOT)/lib/libfftw3_threads.a $(FFTROOT)/lib/libfftw3.a
SRC_MPI =
SRC_FFT = zfftifc_fftw.f90
SRC_libxc = libxcifc_stub.f90
SRC_OBLAS = oblas_stub.f90
SRC_BLIS = blis_stub.f90

LIB_W90 = libwannier.a
SRC_W90S =

Thanks for any advice.

-Sabrina

[J. K. Dewhurst]

Hi Sabrina,

This seems to a be a bug in Intel's compiler. Fortunately it only occurs after the code has finished running, so it can be safely ignored.

Regards,
Kay.

[S. David]

Hi Kay,

Thank you. Do you know why the same error (forrtl: severe (174): SIGSEGV, segmentation fault occurred) might occur a few loops into this input file? It was made for and seemed correct in a very early version of the code (1.4.2) and ran without major error, but different results, in v4.3.6. I was hoping to optimize it in the latest version.

Thank you,
Sabrina

tasks
0
10
20

nempty
120

rgkmax
4.0

gmaxvr
18

trimvg
.true.

vkloff
0.0 0.0 0.0

scissor
0.031

swidth
0.016

beta0
0.01

optcomp
1 1 2

dos
600 100 5 : nwdos, ngrdos, nsmdos
-0.15 0.30 : wdos

ngridk
1 9 1

plot1d
2 100
0.602564 0.000000 0.000000
0.602564 1.000000 0.000000

avec
13.0 0.000000000000 0.0000000
-0.5 0.866025403784 0.0000000
0.0 0.000000000000 9.5695067

scale
5.89972499

sppath
'/projects/elk-7.1.14/species/'

atoms
2 : nspecies
'S.in' : spfname
12 : natoms
0.410256410256410 0.66666666 0.84913998
0.487179487179487 0.66666666 0.84913998
0.564102564102564 0.66666666 0.84913998
0.641025641025641 0.66666666 0.84913998
0.717948717948718 0.66666666 0.84913998
0.794871794871795 0.66666666 0.84913998
0.410256410256410 0.66666666 0.95264741
0.487179487179487 0.66666666 0.95264741
0.564102564102564 0.66666666 0.95264741
0.641025641025641 0.66666666 0.95264741
0.717948717948718 0.66666666 0.95264741
0.794871794871795 0.66666666 0.95264741

'Mo.in' : spfname
6 : natoms
0.435897435897436 0.33333333 0.90089369
0.512820512820513 0.33333333 0.90089369
0.589743589743590 0.33333333 0.90089369
0.666666666666667 0.33333333 0.90089369
0.743589743589744 0.33333333 0.90089369
0.820512820512820 0.33333333 0.90089369

[J. K. Dewhurst]

Hi Sabrina,

It looks like you're running out of OpenMP stack space. You can set this with:

export OMP_STACKSIZE=256M

which should fix the problem.

I'll see if I can get the code to set this by default for the next release.

Regards,
Kay.