Dear all,
I am running the Fe_LDAU test of the 2010 siesta-ldau distribution,
compiling with intel compiler and mpiifort.
I get a deallocation memory problem of the type:
LDAU: Adding Hamiltonian contribution
forrtl: severe (173): A pointer passed to DEALLOCATE points to an array that
cannot be deallocated
Image PC Routine Line Source
siesta 00000000014340CA Unknown Unknown Unknown
siesta 0000000001432BC6 Unknown Unknown Unknown
siesta 0000000001363030 Unknown Unknown Unknown
siesta 000000000130C3BF Unknown Unknown Unknown
siesta 0000000001341630 Unknown Unknown Unknown
siesta 0000000000F29587 alloc_mp_dealloc_ 1357 alloc.F90
siesta 00000000011F32CB m_ldau_mp_hubbard 758 ldau.F
siesta 0000000000D7A3FE m_setup_hamiltoni 199
setup_hamiltonian.F
siesta 0000000000DF2729 m_siesta_forces_m 90
siesta_forces.F
siesta 0000000000DF2AC5 MAIN__ 23 siesta.F
siesta 00000000004084CC Unknown Unknown Unknown
libc.so.6 00007FC89512D76D Unknown Unknown Unknown
siesta 00000000004083C9 Unknown Unknown Unknown
where the problem arises in ldau.F inside subroutine hubbard_term, when the
following instructions are executed
(the problematic line appears to be the one deallocating Vi, but the issue
seems to belong to all deallocations)
call de_alloc( grSki, name='grSki' )
call de_alloc( Ski, name='Ski' )
call de_alloc( xno, name='xno' )
call de_alloc( iono, name='iono' )
call de_alloc( iano, name='iano' )
call de_alloc( listedall, name='listedall' )
call de_alloc( listed, name='listed' )
call de_alloc( Vi, name='Vi' )
call de_alloc( Di, name='Di' )
Do you think this is just a compiler problem? With which compiler does this
issue disappear?
Thank you!
Giovanni
P.S. the problem is idependent of the number of processors I use for parallel
running
