BAD DM NORMALIZATION

I'm trying to run Si2744 system under parallel version. Siesta could complete the job with np=200,320,400,640 but error occurred when I set np as 80,100,160,480 and many other numbers.

Here is my complier options
# arch.make file for gfortran compiler.
# To use this arch.make file you should rename it to
# arch.make
# or make a sym-link.
# For an explanation of the flags see DOCUMENTED-TEMPLATE.make

.SUFFIXES:
.SUFFIXES: .f .F .o .c .a .f90 .F90

SIESTA_ARCH = unknown

FPP = $(FC) -E -P
FC_SERIAL = ifort
CC = mpicc
FC = mpifort
# or mpif90
MPI_INTERFACE = libmpi_f90.a
MPI_INCLUDE = .

FFLAGS = -O2 -fPIC -fp-model source

AR = ar
RANLIB = ranlib

SYS = nag

SP_KIND = 4
DP_KIND = 8
KINDS = $(SP_KIND) $(DP_KIND)

LDFLAGS =

COMP_LIBS = libsiestaLAPACK.a libsiestaBLAS.a

FPPFLAGS = $(DEFS_PREFIX)-DFC_HAVE_ABORT

FPPFLAGS += -DMPI
#MKL
LIBS += -L/opt/intel/2019.update5/mkl/lib/intel64 -lmkl_scalapack_lp64 -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -lmkl_blacs_openmpi_lp64 -lpthread -lm -ldl

# Dependency rules ---------

FFLAGS_DEBUG = -g -O1 -fp-model source # your appropriate flags here...

# The atom.f code is very vulnerable. Particularly the Intel compiler
# will make an erroneous compilation of atom.f with high optimization
# levels.
atom.o: atom.F
 $(FC) -c $(FFLAGS_DEBUG) $(INCFLAGS) $(FPPFLAGS) $(FPPFLAGS_fixed_F) $<
state_analysis.o: state_analysis.F
 $(FC) -c $(FFLAGS_DEBUG) $(INCFLAGS) $(FPPFLAGS) $(FPPFLAGS_fixed_F) $<

.c.o:
 $(CC) -c $(CFLAGS) $(INCFLAGS) $(CPPFLAGS) $<
.F.o:
 $(FC) -c $(FFLAGS) $(INCFLAGS) $(FPPFLAGS) $(FPPFLAGS_fixed_F) $<
.F90.o:
 $(FC) -c $(FFLAGS) $(INCFLAGS) $(FPPFLAGS) $(FPPFLAGS_free_F90) $<
.f.o:
 $(FC) -c $(FFLAGS) $(INCFLAGS) $(FCFLAGS_fixed_f) $<
.f90.o:
 $(FC) -c $(FFLAGS) $(INCFLAGS) $(FCFLAGS_free_f90) $<
And input file
1 SystemName Bulk Silicon
  2 SystemLabel si
  3 NumberOfSpecies 1
  4 %block ChemicalSpeciesLabel
  5 1 14 Si
  6 %endblock ChemicalSpeciesLabel
  7
  8 %include scoord.fdf
  9
 10 xc.functional GGA # Default value
 11 xc.authors PBE # Default value
 12 PAO.BasisSize DZP
 13 ElectronicTemperature 25 meV
 14
 15 MeshCutoff 50. Ry # Equivalent planewave cutoff for the grid
 16 MaxSCFIterations 50 # Maximum number of SCF iterations per step
 17 DM.MixingWeight 0.2 # New DM amount for next SCF cycle
 18 DM.Tolerance 1.d-4 # Tolerance in maximum difference
 19 # between input and output DM
 20 DM.NumberPulay 6 # Number of SCF steps between pulay mixing
 21
 22 # Eigenvalue problem: order-N or diagonalization
 23
 24 SolutionMethod diagon # OrderN or Diagon
 25
 26
 27 MD.TypeOfRun cg # Type of dynamics:
 28 MD.NumCGsteps 0
 29 #MD.VariableCell .true.
 30 MD.MaxForceTol 0.04 eV/Ang
 31 WriteMDXmol .true.
 32 WriteForces .true.
 33 WriteCoorStep
 34
 35 #BandLinesScale pi/a
36 #%block BandLines
 37 #1 1.5 1.5 0.0 K
 38 #38 0.0 0.0 0.0 \Gamma
 39 #36 0.0 2.0 0.0 X
 40 #18 1.0 2.0 0.0 W
 41 #26 1.0 1.0 1.0 L
 42 #31 0.0 0.0 0.0 \Gamma
 43 #%endblock BandLines
 44
 45 # Molecular dynamics and relaxations
and details of error
1 Bad DM normalization: Qtot, Tr[D*S] = 10976.00000000 11098.94423425
   2 Stopping Program from Node: 0
   3 Bad DM normalization: Qtot, Tr[D*S] = 10976.00000000 11098.94423425
   4 Stopping Program from Node: 24

but sometimes
--------------------------------------------------------------------------
   2 Primary job terminated normally, but 1 process returned
   3 a non-zero exit code. Per user-direction, the job has been aborted.
   4 --------------------------------------------------------------------------
   5 forrtl: error (78): process killed (SIGTERM)
   6 Image PC Routine Line Source
   7 siesta 0000000000D1E9C4 for__signal_handl Unknown Unknown