223031012
/
DAMASK_EICMD
mirror of https://github.com/achalhp/DAMASK_EICMD.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

use MPI_f08 if possible 

most PETSc installations provide outdated MPI (f90 version)

MPI_COMM_WORLD is now of derived type (Fortran 08 style)
PETSC_COMM_WORLD is the plain integer (f90 style) alias.
Note that HDF5 is assumed to have f90 interfaces
This commit is contained in:
Martin Diehl 2021-07-08 15:21:35 +02:00
parent 4c3ff4bef4
commit 139f2c177a
14 changed files with 155 additions and 95 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
src/HDF5_utilities.f90
View File

@ -8,6 +8,9 @@ module HDF5_utilities
use HDF5
#ifdef PETSC
use PETSc
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI
#endif
#endif
use prec

12
src/grid/DAMASK_grid.f90
View File

@ -8,7 +8,11 @@
!--------------------------------------------------------------------------------------------------
program DAMASK_grid
#include <petsc/finclude/petscsys.h>
use PETScsys
use PETScSys
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use parallelization
use DAMASK_interface
@ -432,7 +436,7 @@ program DAMASK_grid
print'(/,a,i0,a)', ' increment ', totalIncsCounter, ' NOT converged'
endif; flush(IO_STDOUT)
call MPI_Allreduce(interface_SIGUSR1,signal,1,MPI_LOGICAL,MPI_LOR,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(interface_SIGUSR1,signal,1,MPI_LOGICAL,MPI_LOR,MPI_COMM_WORLD,ierr)
if (ierr /= 0) error stop 'MPI error'
if (mod(inc,loadCases(l)%f_out) == 0 .or. signal) then
print'(1/,a)', ' ... writing results to file ......................................'
@ -440,14 +444,14 @@ program DAMASK_grid
call CPFEM_results(totalIncsCounter,time)
endif
if(signal) call interface_setSIGUSR1(.false.)
call MPI_Allreduce(interface_SIGUSR2,signal,1,MPI_LOGICAL,MPI_LOR,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(interface_SIGUSR2,signal,1,MPI_LOGICAL,MPI_LOR,MPI_COMM_WORLD,ierr)
if (ierr /= 0) error stop 'MPI error'
if (mod(inc,loadCases(l)%f_restart) == 0 .or. signal) then
call mechanical_restartWrite
call CPFEM_restartWrite
endif
if(signal) call interface_setSIGUSR2(.false.)
call MPI_Allreduce(interface_SIGTERM,signal,1,MPI_LOGICAL,MPI_LOR,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(interface_SIGTERM,signal,1,MPI_LOGICAL,MPI_LOR,MPI_COMM_WORLD,ierr)
if (ierr /= 0) error stop 'MPI error'
if (signal) exit loadCaseLooping
endif skipping

17
src/grid/discretization_grid.f90
View File

@ -6,7 +6,10 @@
!--------------------------------------------------------------------------------------------------
module discretization_grid
#include <petsc/finclude/petscsys.h>
use PETScsys
use PETScSys
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use parallelization
@ -75,12 +78,12 @@ subroutine discretization_grid_init(restart)
endif
call MPI_Bcast(grid,3,MPI_INTEGER,0,PETSC_COMM_WORLD, ierr)
call MPI_Bcast(grid,3,MPI_INTEGER,0,MPI_COMM_WORLD, ierr)
if (ierr /= 0) error stop 'MPI error'
if (grid(1) < 2) call IO_error(844, ext_msg='cells(1) must be larger than 1')
call MPI_Bcast(geomSize,3,MPI_DOUBLE,0,PETSC_COMM_WORLD, ierr)
call MPI_Bcast(geomSize,3,MPI_DOUBLE,0,MPI_COMM_WORLD, ierr)
if (ierr /= 0) error stop 'MPI error'
call MPI_Bcast(origin,3,MPI_DOUBLE,0,PETSC_COMM_WORLD, ierr)
call MPI_Bcast(origin,3,MPI_DOUBLE,0,MPI_COMM_WORLD, ierr)
if (ierr /= 0) error stop 'MPI error'
print'(/,a,3(i12 ))', ' cells a b c: ', grid
@ -105,13 +108,13 @@ subroutine discretization_grid_init(restart)
myGrid = [grid(1:2),grid3]
mySize = [geomSize(1:2),size3]
call MPI_Gather(product(grid(1:2))*grid3Offset,1,MPI_INTEGER,displs, 1,MPI_INTEGER,0,PETSC_COMM_WORLD,ierr)
call MPI_Gather(product(grid(1:2))*grid3Offset,1,MPI_INTEGER,displs, 1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
if (ierr /= 0) error stop 'MPI error'
call MPI_Gather(product(myGrid), 1,MPI_INTEGER,sendcounts,1,MPI_INTEGER,0,PETSC_COMM_WORLD,ierr)
call MPI_Gather(product(myGrid), 1,MPI_INTEGER,sendcounts,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
if (ierr /= 0) error stop 'MPI error'
allocate(materialAt(product(myGrid)))
call MPI_scatterv(materialAt_global,sendcounts,displs,MPI_INTEGER,materialAt,size(materialAt),MPI_INTEGER,0,PETSC_COMM_WORLD,ierr)
call MPI_Scatterv(materialAt_global,sendcounts,displs,MPI_INTEGER,materialAt,size(materialAt),MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
if (ierr /= 0) error stop 'MPI error'
call discretization_init(materialAt, &

17
src/grid/grid_damage_spectral.f90
View File

@ -7,8 +7,11 @@
module grid_damage_spectral
#include <petsc/finclude/petscsnes.h>
#include <petsc/finclude/petscdmda.h>
use PETScdmda
use PETScsnes
use PETScDMDA
use PETScSNES
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use parallelization
@ -107,7 +110,7 @@ subroutine grid_damage_spectral_init()
call SNESSetOptionsPrefix(damage_snes,'damage_',ierr);CHKERRQ(ierr)
localK = 0
localK(worldrank) = grid3
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,MPI_COMM_WORLD,ierr)
call DMDACreate3D(PETSC_COMM_WORLD, &
DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, & ! cut off stencil at boundary
DMDA_STENCIL_BOX, & ! Moore (26) neighborhood around central point
@ -187,8 +190,8 @@ function grid_damage_spectral_solution(timeinc) result(solution)
endif
stagNorm = maxval(abs(phi_current - phi_stagInc))
solnNorm = maxval(abs(phi_current))
call MPI_Allreduce(MPI_IN_PLACE,stagNorm,1,MPI_DOUBLE,MPI_MAX,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,solnNorm,1,MPI_DOUBLE,MPI_MAX,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,stagNorm,1,MPI_DOUBLE,MPI_MAX,MPI_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,solnNorm,1,MPI_DOUBLE,MPI_MAX,MPI_COMM_WORLD,ierr)
phi_stagInc = phi_current
solution%stagConverged = stagNorm < max(num%eps_damage_atol, num%eps_damage_rtol*solnNorm)
@ -320,9 +323,9 @@ subroutine updateReference()
enddo
K_ref = K_ref*wgt
call MPI_Allreduce(MPI_IN_PLACE,K_ref,9,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,K_ref,9,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
mu_ref = mu_ref*wgt
call MPI_Allreduce(MPI_IN_PLACE,mu_ref,1,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,mu_ref,1,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
end subroutine updateReference

23
src/grid/grid_mech_FEM.f90
View File

@ -7,8 +7,11 @@
module grid_mechanical_FEM
#include <petsc/finclude/petscsnes.h>
#include <petsc/finclude/petscdmda.h>
use PETScdmda
use PETScsnes
use PETScDMDA
use PETScSNES
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use parallelization
@ -163,7 +166,7 @@ subroutine grid_mechanical_FEM_init
CHKERRQ(ierr)
localK = 0
localK(worldrank) = grid3
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,MPI_COMM_WORLD,ierr)
call DMDACreate3d(PETSC_COMM_WORLD, &
DM_BOUNDARY_PERIODIC, DM_BOUNDARY_PERIODIC, DM_BOUNDARY_PERIODIC, &
DMDA_STENCIL_BOX, &
@ -237,16 +240,16 @@ subroutine grid_mechanical_FEM_init
groupHandle = HDF5_openGroup(fileHandle,'solver')
call HDF5_read(P_aim,groupHandle,'P_aim',.false.)
call MPI_Bcast(P_aim,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(P_aim,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aim,groupHandle,'F_aim',.false.)
call MPI_Bcast(F_aim,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aim,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aim_lastInc,groupHandle,'F_aim_lastInc',.false.)
call MPI_Bcast(F_aim_lastInc,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aim_lastInc,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aimDot,groupHandle,'F_aimDot',.false.)
call MPI_Bcast(F_aimDot,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aimDot,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F,groupHandle,'F')
call HDF5_read(F_lastInc,groupHandle,'F_lastInc')
@ -271,10 +274,10 @@ subroutine grid_mechanical_FEM_init
restartRead2: if (interface_restartInc > 0) then
print'(a,i0,a)', ' reading more restart data of increment ', interface_restartInc, ' from file'
call HDF5_read(C_volAvg,groupHandle,'C_volAvg',.false.)
call MPI_Bcast(C_volAvg,81,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(C_volAvg,81,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(C_volAvgLastInc,groupHandle,'C_volAvgLastInc',.false.)
call MPI_Bcast(C_volAvgLastInc,81,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(C_volAvgLastInc,81,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_closeGroup(groupHandle)
@ -568,7 +571,7 @@ subroutine formResidual(da_local,x_local, &
call utilities_constitutiveResponse(P_current,&
P_av,C_volAvg,devNull, &
F,params%timeinc,params%rotation_BC)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,MPI_COMM_WORLD,ierr)
!--------------------------------------------------------------------------------------------------
! stress BC handling

31
src/grid/grid_mech_spectral_basic.f90
View File

@ -7,8 +7,11 @@
module grid_mechanical_spectral_basic
#include <petsc/finclude/petscsnes.h>
#include <petsc/finclude/petscdmda.h>
use PETScdmda
use PETScsnes
use PETScDMDA
use PETScSNES
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use parallelization
@ -99,7 +102,11 @@ subroutine grid_mechanical_spectral_basic_init
F ! pointer to solution data
PetscInt, dimension(0:worldsize-1) :: localK
integer(HID_T) :: fileHandle, groupHandle
integer :: fileUnit
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
type(MPI_File) :: fileUnit
#else
integer :: fileUnit
#endif
class (tNode), pointer :: &
num_grid, &
debug_grid
@ -154,7 +161,7 @@ subroutine grid_mechanical_spectral_basic_init
call SNESSetOptionsPrefix(snes,'mechanical_',ierr);CHKERRQ(ierr)
localK = 0
localK(worldrank) = grid3
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,MPI_COMM_WORLD,ierr)
call DMDACreate3d(PETSC_COMM_WORLD, &
DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, & ! cut off stencil at boundary
DMDA_STENCIL_BOX, & ! Moore (26) neighborhood around central point
@ -185,16 +192,16 @@ subroutine grid_mechanical_spectral_basic_init
groupHandle = HDF5_openGroup(fileHandle,'solver')
call HDF5_read(P_aim,groupHandle,'P_aim',.false.)
call MPI_Bcast(P_aim,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(P_aim,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aim,groupHandle,'F_aim',.false.)
call MPI_Bcast(F_aim,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aim,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aim_lastInc,groupHandle,'F_aim_lastInc',.false.)
call MPI_Bcast(F_aim_lastInc,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aim_lastInc,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aimDot,groupHandle,'F_aimDot',.false.)
call MPI_Bcast(F_aimDot,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aimDot,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F,groupHandle,'F')
call HDF5_read(F_lastInc,groupHandle,'F_lastInc')
@ -214,16 +221,16 @@ subroutine grid_mechanical_spectral_basic_init
restartRead2: if (interface_restartInc > 0) then
print'(a,i0,a)', ' reading more restart data of increment ', interface_restartInc, ' from file'
call HDF5_read(C_volAvg,groupHandle,'C_volAvg',.false.)
call MPI_Bcast(C_volAvg,81,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(C_volAvg,81,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(C_volAvgLastInc,groupHandle,'C_volAvgLastInc',.false.)
call MPI_Bcast(C_volAvgLastInc,81,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(C_volAvgLastInc,81,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_closeGroup(groupHandle)
call HDF5_closeFile(fileHandle)
call MPI_File_open(PETSC_COMM_WORLD, trim(getSolverJobName())//'.C_ref', &
call MPI_File_open(MPI_COMM_WORLD, trim(getSolverJobName())//'.C_ref', &
MPI_MODE_RDONLY,MPI_INFO_NULL,fileUnit,ierr)
call MPI_File_read(fileUnit,C_minMaxAvg,81,MPI_DOUBLE,MPI_STATUS_IGNORE,ierr)
call MPI_File_close(fileUnit,ierr)
@ -488,7 +495,7 @@ subroutine formResidual(in, F, &
call utilities_constitutiveResponse(residuum, & ! "residuum" gets field of first PK stress (to save memory)
P_av,C_volAvg,C_minMaxAvg, &
F,params%timeinc,params%rotation_BC)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,MPI_COMM_WORLD,ierr)
!--------------------------------------------------------------------------------------------------
! stress BC handling

33
src/grid/grid_mech_spectral_polarisation.f90
View File

@ -7,8 +7,11 @@
module grid_mechanical_spectral_polarisation
#include <petsc/finclude/petscsnes.h>
#include <petsc/finclude/petscdmda.h>
use PETScdmda
use PETScsnes
use PETScDMDA
use PETScSNES
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use parallelization
@ -112,7 +115,11 @@ subroutine grid_mechanical_spectral_polarisation_init
F_tau ! specific (sub)pointer
PetscInt, dimension(0:worldsize-1) :: localK
integer(HID_T) :: fileHandle, groupHandle
integer :: fileUnit
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
type(MPI_File) :: fileUnit
#else
integer :: fileUnit
#endif
class (tNode), pointer :: &
num_grid, &
debug_grid
@ -174,7 +181,7 @@ subroutine grid_mechanical_spectral_polarisation_init
call SNESSetOptionsPrefix(snes,'mechanical_',ierr);CHKERRQ(ierr)
localK = 0
localK(worldrank) = grid3
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,MPI_COMM_WORLD,ierr)
call DMDACreate3d(PETSC_COMM_WORLD, &
DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, & ! cut off stencil at boundary
DMDA_STENCIL_BOX, & ! Moore (26) neighborhood around central point
@ -207,16 +214,16 @@ subroutine grid_mechanical_spectral_polarisation_init
groupHandle = HDF5_openGroup(fileHandle,'solver')
call HDF5_read(P_aim,groupHandle,'P_aim',.false.)
call MPI_Bcast(P_aim,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(P_aim,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aim,groupHandle,'F_aim',.false.)
call MPI_Bcast(F_aim,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aim,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aim_lastInc,groupHandle,'F_aim_lastInc',.false.)
call MPI_Bcast(F_aim_lastInc,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aim_lastInc,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F_aimDot,groupHandle,'F_aimDot',.false.)
call MPI_Bcast(F_aimDot,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(F_aimDot,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(F,groupHandle,'F')
call HDF5_read(F_lastInc,groupHandle,'F_lastInc')
@ -240,16 +247,16 @@ subroutine grid_mechanical_spectral_polarisation_init
restartRead2: if (interface_restartInc > 0) then
print'(a,i0,a)', ' reading more restart data of increment ', interface_restartInc, ' from file'
call HDF5_read(C_volAvg,groupHandle,'C_volAvg',.false.)
call MPI_Bcast(C_volAvg,81,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(C_volAvg,81,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_read(C_volAvgLastInc,groupHandle,'C_volAvgLastInc',.false.)
call MPI_Bcast(C_volAvgLastInc,81,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(C_volAvgLastInc,81,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
call HDF5_closeGroup(groupHandle)
call HDF5_closeFile(fileHandle)
call MPI_File_open(PETSC_COMM_WORLD, trim(getSolverJobName())//'.C_ref', &
call MPI_File_open(MPI_COMM_WORLD, trim(getSolverJobName())//'.C_ref', &
MPI_MODE_RDONLY,MPI_INFO_NULL,fileUnit,ierr)
call MPI_File_read(fileUnit,C_minMaxAvg,81,MPI_DOUBLE,MPI_STATUS_IGNORE,ierr)
call MPI_File_close(fileUnit,ierr)
@ -544,7 +551,7 @@ subroutine formResidual(in, FandF_tau, &
X_RANGE, Y_RANGE, Z_RANGE)
F_av = sum(sum(sum(F,dim=5),dim=4),dim=3) * wgt
call MPI_Allreduce(MPI_IN_PLACE,F_av,9,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,F_av,9,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
call SNESGetNumberFunctionEvals(snes,nfuncs,ierr); CHKERRQ(ierr)
call SNESGetIterationNumber(snes,PETScIter,ierr); CHKERRQ(ierr)
@ -588,7 +595,7 @@ subroutine formResidual(in, FandF_tau, &
call utilities_constitutiveResponse(residual_F, & ! "residuum" gets field of first PK stress (to save memory)
P_av,C_volAvg,C_minMaxAvg, &
F - residual_F_tau/num%beta,params%timeinc,params%rotation_BC)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,MPI_COMM_WORLD,ierr)
!--------------------------------------------------------------------------------------------------
! stress BC handling

17
src/grid/grid_thermal_spectral.f90
View File

@ -7,8 +7,11 @@
module grid_thermal_spectral
#include <petsc/finclude/petscsnes.h>
#include <petsc/finclude/petscdmda.h>
use PETScdmda
use PETScsnes
use PETScDMDA
use PETScSNES
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use parallelization
@ -102,7 +105,7 @@ subroutine grid_thermal_spectral_init(T_0)
call SNESSetOptionsPrefix(thermal_snes,'thermal_',ierr);CHKERRQ(ierr)
localK = 0
localK(worldrank) = grid3
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,localK,worldsize,MPI_INTEGER,MPI_SUM,MPI_COMM_WORLD,ierr)
call DMDACreate3D(PETSC_COMM_WORLD, &
DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, DM_BOUNDARY_NONE, & ! cut off stencil at boundary
DMDA_STENCIL_BOX, & ! Moore (26) neighborhood around central point
@ -182,8 +185,8 @@ function grid_thermal_spectral_solution(timeinc) result(solution)
endif
stagNorm = maxval(abs(T_current - T_stagInc))
solnNorm = maxval(abs(T_current))
call MPI_Allreduce(MPI_IN_PLACE,stagNorm,1,MPI_DOUBLE,MPI_MAX,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,solnNorm,1,MPI_DOUBLE,MPI_MAX,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,stagNorm,1,MPI_DOUBLE,MPI_MAX,MPI_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,solnNorm,1,MPI_DOUBLE,MPI_MAX,MPI_COMM_WORLD,ierr)
T_stagInc = T_current
solution%stagConverged = stagNorm < max(num%eps_thermal_atol, num%eps_thermal_rtol*solnNorm)
@ -310,9 +313,9 @@ subroutine updateReference()
enddo
K_ref = K_ref*wgt
call MPI_Allreduce(MPI_IN_PLACE,K_ref,9,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,K_ref,9,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
mu_ref = mu_ref*wgt
call MPI_Allreduce(MPI_IN_PLACE,mu_ref,1,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,mu_ref,1,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
end subroutine updateReference

40
src/grid/spectral_utilities.f90
View File

@ -8,6 +8,9 @@ module spectral_utilities
#include <petsc/finclude/petscsys.h>
use PETScSys
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use DAMASK_interface
@ -591,7 +594,7 @@ real(pReal) function utilities_divergenceRMS()
conjg(-xi1st(1:3,grid1Red,j,k))*rescaledGeom))**2.0_pReal)
enddo; enddo
if(grid(1) == 1) utilities_divergenceRMS = utilities_divergenceRMS * 0.5_pReal ! counted twice in case of grid(1) == 1
call MPI_Allreduce(MPI_IN_PLACE,utilities_divergenceRMS,1,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,utilities_divergenceRMS,1,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
utilities_divergenceRMS = sqrt(utilities_divergenceRMS) * wgt ! RMS in real space calculated with Parsevals theorem from Fourier space
@ -651,7 +654,7 @@ real(pReal) function utilities_curlRMS()
+ sum(real(curl_fourier)**2.0_pReal + aimag(curl_fourier)**2.0_pReal) ! this layer (Nyquist) does not have a conjugate complex counterpart (if grid(1) /= 1)
enddo; enddo
call MPI_Allreduce(MPI_IN_PLACE,utilities_curlRMS,1,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,utilities_curlRMS,1,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
utilities_curlRMS = sqrt(utilities_curlRMS) * wgt
if(grid(1) == 1) utilities_curlRMS = utilities_curlRMS * 0.5_pReal ! counted twice in case of grid(1) == 1
@ -816,7 +819,7 @@ subroutine utilities_constitutiveResponse(P,P_av,C_volAvg,C_minmaxAvg,&
P = reshape(homogenization_P, [3,3,grid(1),grid(2),grid3])
P_av = sum(sum(sum(P,dim=5),dim=4),dim=3) * wgt ! average of P
call MPI_Allreduce(MPI_IN_PLACE,P_av,9,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,P_av,9,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
if (debugRotation) print'(/,a,/,2(3(2x,f12.4,1x)/),3(2x,f12.4,1x))', &
' Piola--Kirchhoff stress (lab) / MPa =', transpose(P_av)*1.e-6_pReal
if(present(rotation_BC)) P_av = rotation_BC%rotate(P_av)
@ -840,21 +843,21 @@ subroutine utilities_constitutiveResponse(P,P_av,C_volAvg,C_minmaxAvg,&
end do
valueAndRank = [dPdF_norm_max,real(worldrank,pReal)]
call MPI_Allreduce(MPI_IN_PLACE,valueAndRank,1, MPI_2DOUBLE_PRECISION, MPI_MAXLOC, PETSC_COMM_WORLD, ierr)
call MPI_Allreduce(MPI_IN_PLACE,valueAndRank,1, MPI_2DOUBLE_PRECISION, MPI_MAXLOC, MPI_COMM_WORLD, ierr)
if (ierr /= 0) error stop 'MPI error'
call MPI_Bcast(dPdF_max,81,MPI_DOUBLE,int(valueAndRank(2)),PETSC_COMM_WORLD, ierr)
call MPI_Bcast(dPdF_max,81,MPI_DOUBLE,int(valueAndRank(2)),MPI_COMM_WORLD, ierr)
if (ierr /= 0) error stop 'MPI error'
valueAndRank = [dPdF_norm_min,real(worldrank,pReal)]
call MPI_Allreduce(MPI_IN_PLACE,valueAndRank,1, MPI_2DOUBLE_PRECISION, MPI_MINLOC, PETSC_COMM_WORLD, ierr)
call MPI_Allreduce(MPI_IN_PLACE,valueAndRank,1, MPI_2DOUBLE_PRECISION, MPI_MINLOC, MPI_COMM_WORLD, ierr)
if (ierr /= 0) error stop 'MPI error'
call MPI_Bcast(dPdF_min,81,MPI_DOUBLE,int(valueAndRank(2)),PETSC_COMM_WORLD, ierr)
call MPI_Bcast(dPdF_min,81,MPI_DOUBLE,int(valueAndRank(2)),MPI_COMM_WORLD, ierr)
if (ierr /= 0) error stop 'MPI error'
C_minmaxAvg = 0.5_pReal*(dPdF_max + dPdF_min)
C_volAvg = sum(homogenization_dPdF,dim=5)
call MPI_Allreduce(MPI_IN_PLACE,C_volAvg,81,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,C_volAvg,81,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
if (ierr /= 0) error stop 'MPI error'
C_volAvg = C_volAvg * wgt
@ -909,7 +912,7 @@ function utilities_forwardField(timeinc,field_lastInc,rate,aim)
utilities_forwardField = field_lastInc + rate*timeinc
if (present(aim)) then !< correct to match average
fieldDiff = sum(sum(sum(utilities_forwardField,dim=5),dim=4),dim=3)*wgt
call MPI_Allreduce(MPI_IN_PLACE,fieldDiff,9,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,fieldDiff,9,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
fieldDiff = fieldDiff - aim
utilities_forwardField = utilities_forwardField - &
spread(spread(spread(fieldDiff,3,grid(1)),4,grid(2)),5,grid3)
@ -982,8 +985,13 @@ subroutine utilities_updateCoords(F)
rank_t, rank_b, &
c, &
ierr
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
type(MPI_Request), dimension(4) :: request
type(MPI_Status), dimension(4) :: status
#else
integer, dimension(4) :: request
integer, dimension(MPI_STATUS_SIZE,4) :: status
#endif
real(pReal), dimension(3) :: step
real(pReal), dimension(3,3) :: Favg
integer, dimension(3) :: me
@ -1018,7 +1026,7 @@ subroutine utilities_updateCoords(F)
!--------------------------------------------------------------------------------------------------
! average F
if (grid3Offset == 0) Favg = real(tensorField_fourier(1:3,1:3,1,1,1),pReal)*wgt
call MPI_Bcast(Favg,9,MPI_DOUBLE,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(Favg,9,MPI_DOUBLE,0,MPI_COMM_WORLD,ierr)
if(ierr /=0) error stop 'MPI error'
!--------------------------------------------------------------------------------------------------
@ -1029,20 +1037,24 @@ subroutine utilities_updateCoords(F)
rank_b = modulo(worldrank-1,worldsize)
! send bottom layer to process below
call MPI_Isend(IPfluct_padded(:,:,:,2), c,MPI_DOUBLE,rank_b,0,PETSC_COMM_WORLD,request(1),ierr)
call MPI_Isend(IPfluct_padded(:,:,:,2), c,MPI_DOUBLE,rank_b,0,MPI_COMM_WORLD,request(1),ierr)
if(ierr /=0) error stop 'MPI error'
call MPI_Irecv(IPfluct_padded(:,:,:,grid3+2),c,MPI_DOUBLE,rank_t,0,PETSC_COMM_WORLD,request(2),ierr)
call MPI_Irecv(IPfluct_padded(:,:,:,grid3+2),c,MPI_DOUBLE,rank_t,0,MPI_COMM_WORLD,request(2),ierr)
if(ierr /=0) error stop 'MPI error'
! send top layer to process above
call MPI_Isend(IPfluct_padded(:,:,:,grid3+1),c,MPI_DOUBLE,rank_t,1,PETSC_COMM_WORLD,request(3),ierr)
call MPI_Isend(IPfluct_padded(:,:,:,grid3+1),c,MPI_DOUBLE,rank_t,1,MPI_COMM_WORLD,request(3),ierr)
if(ierr /=0) error stop 'MPI error'
call MPI_Irecv(IPfluct_padded(:,:,:,1), c,MPI_DOUBLE,rank_b,1,PETSC_COMM_WORLD,request(4),ierr)
call MPI_Irecv(IPfluct_padded(:,:,:,1), c,MPI_DOUBLE,rank_b,1,MPI_COMM_WORLD,request(4),ierr)
if(ierr /=0) error stop 'MPI error'
call MPI_Waitall(4,request,status,ierr)
if(ierr /=0) error stop 'MPI error'
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
! ToDo
#else
if(any(status(MPI_ERROR,:) /= 0)) error stop 'MPI error'
#endif
!--------------------------------------------------------------------------------------------------
! calculate nodal displacements

13
src/mesh/FEM_utilities.f90
View File

@ -7,10 +7,13 @@ module FEM_utilities
#include <petsc/finclude/petscdmda.h>
#include <petsc/finclude/petscis.h>
use PETScdmplex
use PETScdmda
use PETScis
use PETScDMplex
use PETScDMDA
use PETScIS
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use config
use math
@ -165,7 +168,7 @@ subroutine utilities_constitutiveResponse(timeinc,P_av,forwardData)
cutBack = .false. ! reset cutBack status
P_av = sum(homogenization_P,dim=3) * wgt
call MPI_Allreduce(MPI_IN_PLACE,P_av,9,MPI_DOUBLE,MPI_SUM,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,P_av,9,MPI_DOUBLE,MPI_SUM,MPI_COMM_WORLD,ierr)
end subroutine utilities_constitutiveResponse

17
src/mesh/discretization_mesh.f90
View File

@ -8,9 +8,12 @@ module discretization_mesh
#include <petsc/finclude/petscdmplex.h>
#include <petsc/finclude/petscis.h>
#include <petsc/finclude/petscdmda.h>
use PETScdmplex
use PETScdmda
use PETScis
use PETScDMplex
use PETScDMDA
use PETScIS
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use DAMASK_interface
use parallelization
@ -111,9 +114,9 @@ subroutine discretization_mesh_init(restart)
! get number of IDs in face sets (for boundary conditions?)
call DMGetLabelSize(globalMesh,'Face Sets',mesh_Nboundaries,ierr)
CHKERRQ(ierr)
call MPI_Bcast(mesh_Nboundaries,1,MPI_INTEGER,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(mesh_NcpElemsGlobal,1,MPI_INTEGER,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(dimPlex,1,MPI_INTEGER,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(mesh_Nboundaries,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
call MPI_Bcast(mesh_NcpElemsGlobal,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
call MPI_Bcast(dimPlex,1,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
if (worldrank == 0) then
call DMClone(globalMesh,geomMesh,ierr)
@ -134,7 +137,7 @@ subroutine discretization_mesh_init(restart)
CHKERRQ(ierr)
call ISRestoreIndicesF90(faceSetIS,pFaceSets,ierr)
endif
call MPI_Bcast(mesh_boundaries,mesh_Nboundaries,MPI_INTEGER,0,PETSC_COMM_WORLD,ierr)
call MPI_Bcast(mesh_boundaries,mesh_Nboundaries,MPI_INTEGER,0,MPI_COMM_WORLD,ierr)
call DMDestroy(globalMesh,ierr); CHKERRQ(ierr)

7
src/mesh/mesh_mech_FEM.f90
View File

@ -9,10 +9,13 @@ module mesh_mechanical_FEM
#include <petsc/finclude/petscdm.h>
#include <petsc/finclude/petsc.h>
use PETScsnes
use PETScSNES
use PETScDM
use PETScDMplex
use PETScDT
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
use prec
use FEM_utilities
@ -396,7 +399,7 @@ subroutine FEM_mechanical_formResidual(dm_local,xx_local,f_local,dummy,ierr)
!--------------------------------------------------------------------------------------------------
! evaluate constitutive response
call Utilities_constitutiveResponse(params%timeinc,P_av,ForwardData)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,PETSC_COMM_WORLD,ierr)
call MPI_Allreduce(MPI_IN_PLACE,terminallyIll,1,MPI_LOGICAL,MPI_LOR,MPI_COMM_WORLD,ierr)
ForwardData = .false.
!--------------------------------------------------------------------------------------------------

9
src/parallelization.f90
View File

@ -8,7 +8,10 @@ module parallelization
#ifdef PETSC
#include <petsc/finclude/petscsys.h>
use petscsys
use PETScSys
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
!$ use OMP_LIB
#endif
use prec
@ -60,12 +63,12 @@ subroutine parallelization_init
#endif
CHKERRQ(petsc_err)
call MPI_Comm_rank(PETSC_COMM_WORLD,worldrank,err)
call MPI_Comm_rank(MPI_COMM_WORLD,worldrank,err)
if (err /= 0) error stop 'Could not determine worldrank'
if (worldrank == 0) print'(/,a)', ' <<<+- parallelization init -+>>>'
call MPI_Comm_size(PETSC_COMM_WORLD,worldsize,err)
call MPI_Comm_size(MPI_COMM_WORLD,worldsize,err)
if (err /= 0) error stop 'Could not determine worldsize'
if (worldrank == 0) print'(a,i3)', ' MPI processes: ',worldsize

11
src/results.f90
View File

@ -13,6 +13,9 @@ module results
use HDF5
#ifdef PETSC
use PETSc
#ifndef PETSC_HAVE_MPI_F90MODULE_VISIBILITY
use MPI_f08
#endif
#endif
implicit none
@ -461,7 +464,7 @@ subroutine results_mapping_phase(ID,entry,label)
call h5pset_dxpl_mpio_f(plist_id, H5FD_MPIO_COLLECTIVE_F, hdferr)
if(hdferr < 0) error stop 'HDF5 error'
call MPI_allreduce(MPI_IN_PLACE,writeSize,worldsize,MPI_INT,MPI_SUM,PETSC_COMM_WORLD,ierr) ! get output at each process
call MPI_Allreduce(MPI_IN_PLACE,writeSize,worldsize,MPI_INT,MPI_SUM,MPI_COMM_WORLD,ierr) ! get output at each process
if(ierr /= 0) error stop 'MPI error'
entryOffset = 0
@ -470,7 +473,7 @@ subroutine results_mapping_phase(ID,entry,label)
entryOffset(ID(co,ce),worldrank) = entryOffset(ID(co,ce),worldrank) +1
enddo
enddo
call MPI_allreduce(MPI_IN_PLACE,entryOffset,size(entryOffset),MPI_INT,MPI_SUM,PETSC_COMM_WORLD,ierr)! get offset at each process
call MPI_Allreduce(MPI_IN_PLACE,entryOffset,size(entryOffset),MPI_INT,MPI_SUM,MPI_COMM_WORLD,ierr)! get offset at each process
if(ierr /= 0) error stop 'MPI error'
entryOffset(:,worldrank) = sum(entryOffset(:,0:worldrank-1),2)
do co = 1, size(ID,1)
@ -614,14 +617,14 @@ subroutine results_mapping_homogenization(ID,entry,label)
call h5pset_dxpl_mpio_f(plist_id, H5FD_MPIO_COLLECTIVE_F, hdferr)
if(hdferr < 0) error stop 'HDF5 error'
call MPI_allreduce(MPI_IN_PLACE,writeSize,worldsize,MPI_INT,MPI_SUM,PETSC_COMM_WORLD,ierr) ! get output at each process
call MPI_Allreduce(MPI_IN_PLACE,writeSize,worldsize,MPI_INT,MPI_SUM,MPI_COMM_WORLD,ierr) ! get output at each process
if(ierr /= 0) error stop 'MPI error'
entryOffset = 0
do ce = 1, size(ID,1)
entryOffset(ID(ce),worldrank) = entryOffset(ID(ce),worldrank) +1
enddo
call MPI_allreduce(MPI_IN_PLACE,entryOffset,size(entryOffset),MPI_INT,MPI_SUM,PETSC_COMM_WORLD,ierr) ! get offset at each process
call MPI_Allreduce(MPI_IN_PLACE,entryOffset,size(entryOffset),MPI_INT,MPI_SUM,MPI_COMM_WORLD,ierr)! get offset at each process
if(ierr /= 0) error stop 'MPI error'
entryOffset(:,worldrank) = sum(entryOffset(:,0:worldrank-1),2)
do ce = 1, size(ID,1)