OMP bugfix for FPI integrator, memory-efficient RK4
This commit is contained in:
Martin Diehl
parent
9c95ce36f4
commit
e46220cd8a
|
|
@ -1016,7 +1016,7 @@ subroutine integrateStateFPI
|
|||
nonlocalBroken
|
||||
|
||||
nonlocalBroken = .false.
|
||||
!$OMP PARALLEL DO PRIVATE(sizeDotState,r,zeta,p,c,plastic_dotState_p1, plastic_dotState_p2,source_dotState_p1, source_dotState_p2)
|
||||
!$OMP PARALLEL DO PRIVATE(sizeDotState,r,zeta,p,c,plastic_dotState_p1, plastic_dotState_p2,source_dotState)
|
||||
do e = FEsolving_execElem(1),FEsolving_execElem(2)
|
||||
do i = FEsolving_execIP(1),FEsolving_execIP(2)
|
||||
do g = 1,homogenization_Ngrains(material_homogenizationAt(e))
|
||||
|
|
@ -1359,8 +1359,10 @@ subroutine integrateStateRK4
|
|||
logical :: &
|
||||
nonlocalBroken
|
||||
|
||||
real(pReal), dimension(constitutive_source_maxSizeDotState,4,maxval(phase_Nsources)) :: source_RK4dotState
|
||||
real(pReal), dimension(constitutive_plasticity_maxSizeDotState,4) :: plastic_RK4dotState
|
||||
nonlocalBroken = .false.
|
||||
!$OMP PARALLEL DO PRIVATE(sizeDotState,p,c)
|
||||
!$OMP PARALLEL DO PRIVATE(sizeDotState,p,c,source_RK4dotState,plastic_RK4dotState)
|
||||
do e = FEsolving_execElem(1),FEsolving_execElem(2)
|
||||
do i = FEsolving_execIP(1),FEsolving_execIP(2)
|
||||
do g = 1,homogenization_Ngrains(material_homogenizationAt(e))
|
||||
|
|
@ -1382,20 +1384,23 @@ subroutine integrateStateRK4
|
|||
if(.not. crystallite_todo(g,i,e)) cycle
|
||||
|
||||
do stage = 1,3
|
||||
|
||||
plasticState(p)%RK4dotState(stage,:,c) = plasticState(p)%dotState(:,c)
|
||||
plasticState(p)%dotState(:,c) = A(1,stage) * plasticState(p)%RK4dotState(1,:,c)
|
||||
sizeDotState = plasticState(p)%sizeDotState
|
||||
plastic_RK4dotState(1:sizeDotState,stage) = plasticState(p)%dotState(:,c)
|
||||
plasticState(p)%dotState(:,c) = A(1,stage) * plastic_RK4dotState(1:sizeDotState,1)
|
||||
do s = 1, phase_Nsources(p)
|
||||
sourceState(p)%p(s)%RK4dotState(stage,:,c) = sourceState(p)%p(s)%dotState(:,c)
|
||||
sourceState(p)%p(s)%dotState(:,c) = A(1,stage) * sourceState(p)%p(s)%RK4dotState(1,:,c)
|
||||
sizeDotState = sourceState(p)%p(s)%sizeDotState
|
||||
source_RK4dotState(1:sizeDotState,stage,s) = sourceState(p)%p(s)%dotState(:,c)
|
||||
sourceState(p)%p(s)%dotState(:,c) = A(1,stage) * source_RK4dotState(1:sizeDotState,1,s)
|
||||
enddo
|
||||
|
||||
do n = 2, stage
|
||||
sizeDotState = plasticState(p)%sizeDotState
|
||||
plasticState(p)%dotState(:,c) = plasticState(p)%dotState(:,c) &
|
||||
+ A(n,stage) * plasticState(p)%RK4dotState(n,:,c)
|
||||
+ A(n,stage) * plastic_RK4dotState(1:sizeDotState,n)
|
||||
do s = 1, phase_Nsources(p)
|
||||
sizeDotState = sourceState(p)%p(s)%sizeDotState
|
||||
sourceState(p)%p(s)%dotState(:,c) = sourceState(p)%p(s)%dotState(:,c) &
|
||||
+ A(n,stage) * sourceState(p)%p(s)%RK4dotState(n,:,c)
|
||||
+ A(n,stage) * source_RK4dotState(1:sizeDotState,n,s)
|
||||
enddo
|
||||
enddo
|
||||
|
||||
|
|
@ -1438,9 +1443,9 @@ subroutine integrateStateRK4
|
|||
|
||||
sizeDotState = plasticState(p)%sizeDotState
|
||||
|
||||
plasticState(p)%RK4dotState(4,:,c) = plasticState (p)%dotState(:,c)
|
||||
plastic_RK4dotState(1:sizeDotState,4) = plasticState (p)%dotState(:,c)
|
||||
|
||||
plasticState(p)%dotState(:,c) = matmul(B,plasticState(p)%RK4dotState(1:4,1:sizeDotState,c))
|
||||
plasticState(p)%dotState(:,c) = matmul(plastic_RK4dotState(1:sizeDotState,1:4),B)
|
||||
plasticState(p)%state(1:sizeDotState,c) = plasticState(p)%subState0(1:sizeDotState,c) &
|
||||
+ plasticState(p)%dotState (1:sizeDotState,c) &
|
||||
* crystallite_subdt(g,i,e)
|
||||
|
|
@ -1448,9 +1453,9 @@ subroutine integrateStateRK4
|
|||
do s = 1, phase_Nsources(p)
|
||||
sizeDotState = sourceState(p)%p(s)%sizeDotState
|
||||
|
||||
sourceState(p)%p(s)%RK4dotState(4,:,c) = sourceState(p)%p(s)%dotState(:,c)
|
||||
source_RK4dotState(1:sizeDotState,4,s) = sourceState(p)%p(s)%dotState(:,c)
|
||||
|
||||
sourceState(p)%p(s)%dotState(:,c) = matmul(B,sourceState(p)%p(s)%RK4dotState(1:4,1:sizeDotState,c))
|
||||
sourceState(p)%p(s)%dotState(:,c) = matmul(source_RK4dotState(1:sizeDotState,1:4,s),B)
|
||||
sourceState(p)%p(s)%state(1:sizeDotState,c) = sourceState(p)%p(s)%subState0(1:sizeDotState,c) &
|
||||
+ sourceState(p)%p(s)%dotState (1:sizeDotState,c) &
|
||||
* crystallite_subdt(g,i,e)
|
||||
|
|
|
|||
|
|
@ -724,8 +724,6 @@ subroutine material_allocatePlasticState(phase,NipcMyPhase,&
|
|||
allocate(plasticState(phase)%state (sizeState,NipcMyPhase), source=0.0_pReal)
|
||||
|
||||
allocate(plasticState(phase)%dotState (sizeDotState,NipcMyPhase),source=0.0_pReal)
|
||||
if (numerics_integrator == 4) &
|
||||
allocate(plasticState(phase)%RK4dotState (4,sizeDotState,NipcMyPhase),source=0.0_pReal)
|
||||
if (numerics_integrator == 5) &
|
||||
allocate(plasticState(phase)%RKCK45dotState (6,sizeDotState,NipcMyPhase),source=0.0_pReal)
|
||||
|
||||
|
|
@ -758,8 +756,6 @@ subroutine material_allocateSourceState(phase,of,NipcMyPhase,&
|
|||
allocate(sourceState(phase)%p(of)%state (sizeState,NipcMyPhase), source=0.0_pReal)
|
||||
|
||||
allocate(sourceState(phase)%p(of)%dotState (sizeDotState,NipcMyPhase),source=0.0_pReal)
|
||||
if (numerics_integrator == 4) &
|
||||
allocate(sourceState(phase)%p(of)%RK4dotState (4,sizeDotState,NipcMyPhase),source=0.0_pReal)
|
||||
if (numerics_integrator == 5) &
|
||||
allocate(sourceState(phase)%p(of)%RKCK45dotState (6,sizeDotState,NipcMyPhase),source=0.0_pReal)
|
||||
|
||||
|
|
|
|||
|
|
@ -46,11 +46,8 @@ module prec
|
|||
deltaState !< increment of state change
|
||||
real(pReal), allocatable, dimension(:,:) :: &
|
||||
partionedState0, &
|
||||
subState0, &
|
||||
previousDotState, &
|
||||
previousDotState2
|
||||
subState0
|
||||
real(pReal), allocatable, dimension(:,:,:) :: &
|
||||
RK4dotState, &
|
||||
RKCK45dotState
|
||||
end type
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue