Namespaces
	detail

Classes
struct	HasInverseMemFn
	Type trait that tests if T has an InverseImage class member function. More...

struct	IsIndexMapping
	Tests if a given type IndexMap is usable as an index mapping between two index based coordinate systems. More...

struct	MultiBlockIndexMapping
	This struct describes an affine index transformation for two coordinate systems. More...

struct	MultiBlockCommMetaData
	This is the index mapping based on the DTOS MultiBlockDestToSrc. More...

struct	IsFabProjection
	This type trait tests if a type P is a projection for FAB. More...

struct	Identity
	This class acts as a default no-op operator. More...

struct	MapComponents
	This class takes a projection and a component map and combines them to form a new projection. More...

struct	SwapComponents
	This is a permutation where only two components are swapped. More...

struct	SwapComponents< I, -1 >

struct	SwapComponents<-1, J >

struct	SwapComponents<-1, -1 >

struct	CommData
	This class holds data buffers for either immediate MPI send or recv calls. More...

struct	CommHandler
	This class stores both recv and send buffers with an associated MPI tag. More...

struct	IsDataPacking
	This type trait tests if a given type DP satisfies the DataPacking concept for type FAB. More...

struct	PackComponents
	Contains information about which components take part of the data transaction. More...

struct	ApplyDtosAndProjectionOnReciever
	This class specializes behaviour on local copies and unpacking receive buffers. More...

struct	NoLocalCopy

struct	DoLocalCopy

Typedefs
template<typename T , typename... Args>
using	Inverse_t = decltype(std::declval< T >().Inverse(std::declval< Args >()...))
	Return type of an InverseImage class member function. More...

using	DynamicSwapComponents = SwapComponents<-1, -1 >

template<typename... Args>
using	PrepareSendBuffers_t = decltype(PrepareSendBuffers(std::declval< Args >()...))

template<typename... Args>
using	PrepareRecvBuffers_t = decltype(PrepareRecvBuffers(std::declval< Args >()...))

template<typename... Args>
using	PackSendBuffers_t = decltype(PackSendBuffers(std::declval< Args >()...))

template<typename... Args>
using	UnpackRecvBuffers_t = decltype(UnpackRecvBuffers(std::declval< Args >()...))

template<typename... Args>
using	LocalCopy_t = decltype(LocalCopy(std::declval< Args >()...))

Functions
void	PrepareCommBuffers (CommData &comm, const FabArrayBase::MapOfCopyComTagContainers &cctc, int n_components, std::size_t object_size, std::size_t align)
	Fill all class member variables of comm but the request and the stats vector. More...

void	PostRecvs (CommData &recv, int mpi_tag)
	Initiate all recvieves with MPI_Irecv calls associated with tag mpi_tag. More...

void	PostSends (CommData &send, int mpi_tag)
	Initiate all sends with MPI_Isend calls associated with tag mpi_tag. More...

template MultiBlockCommMetaData	ParallelCopy (FabArray< FArrayBox > &dest, const Box &destbox, const FabArray< FArrayBox > &src, int destcomp, int srccomp, int numcomp, const IntVect &ngrow, MultiBlockIndexMapping const &, Identity const &)

template<typename DTOS >
std::enable_if_t< IsCallableR< Dim3, DTOS, Dim3 >::value, IntVect >	Apply (DTOS const &dtos, const IntVect &iv)
	Applies the Dim3 to Dim3 mapping onto IntVects. More...

template<typename DTOS >
std::enable_if_t< IsCallableR< Dim3, DTOS, Dim3 >::value &&!IsCallableR< IndexType, DTOS, IndexType >::value, Box >	Image (DTOS const &dtos, const Box &box)
	Applies the Dim3 to Dim3 mapping onto Boxes but does not change the index type. More...

template<typename DTOS >
std::enable_if_t< IsCallableR< Dim3, DTOS, Dim3 >::value &&IsCallableR< IndexType, DTOS, IndexType >::value, Box >	Image (DTOS const &dtos, const Box &box)
	Applies the Dim3 to Dim3 mapping onto Boxes and maps the index type. More...

template<typename DTOS >
std::enable_if_t< HasInverseMemFn< DTOS >::value, IntVect >	ApplyInverse (DTOS const &dtos, const IntVect &iv)
	Applies the Dim3 to Dim3 invserse mapping onto IntVects. More...

template<typename DTOS >
std::enable_if_t< HasInverseMemFn< DTOS >::value &&!IsCallableR< IndexType, DTOS, IndexType >::value, Box >	InverseImage (DTOS const &dtos, const Box &box)
	Applies the inverse Dim3 to Dim3 mapping onto Boxes without changing the index type. More...

template<typename DTOS >
std::enable_if_t< HasInverseMemFn< DTOS >::value &&IsCallableR< IndexType, DTOS, IndexType >::value, Box >	InverseImage (DTOS const &dtos, const Box &box)
	Applies the inverse Dim3 to Dim3 mapping onto Boxes. More...

template<class FAB , class DTOS = Identity, class Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	local_copy_cpu (FabArray< FAB > &dest, const FabArray< FAB > &src, int dcomp, int scomp, int ncomp, FabArrayBase::CopyComTagsContainer const &local_tags, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{}) noexcept

template<class FAB , class DTOS = Identity, class Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	unpack_recv_buffer_cpu (FabArray< FAB > &mf, int dcomp, int ncomp, Vector< char * > const &recv_data, Vector< std::size_t > const &recv_size, Vector< FabArrayBase::CopyComTagsContainer const * > const &recv_cctc, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{}) noexcept

template<class FAB , class DTOS = Identity, class Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	local_copy_gpu (FabArray< FAB > &dest, const FabArray< FAB > &src, int dcomp, int scomp, int ncomp, FabArrayBase::CopyComTagsContainer const &local_tags, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{}) noexcept

template<class FAB , class DTOS = Identity, class Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	unpack_recv_buffer_gpu (FabArray< FAB > &mf, int scomp, int ncomp, Vector< char * > const &recv_data, Vector< std::size_t > const &recv_size, Vector< FabArrayBase::CopyComTagsContainer const * > const &recv_cctc, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{})

template<typename FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	LocalCopy (const PackComponents &components, FabArray< FAB > &dest, const FabArray< FAB > &src, const FabArrayBase::CopyComTagsContainer &local_tags)
	Dispatch local copies to the default behaviour that knows no DTOS nor projection. More...

template<typename FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	PrepareSendBuffers (const PackComponents &components, FabArray< FAB > &dest, const FabArray< FAB > &src, CommData &comm, const FabArrayBase::MapOfCopyComTagContainers &cctc)
	Calls PrepareComBuffers. More...

template<typename FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	PrepareRecvBuffers (const PackComponents &components, FabArray< FAB > &dest, const FabArray< FAB > &src, CommData &comm, const FabArrayBase::MapOfCopyComTagContainers &cctc)
	Calls PrepareComBuffers. More...

template<typename FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	PackSendBuffers (const PackComponents &components, const FabArray< FAB > &src, CommData &send)
	Serializes FAB data without any knowledge of a DTOS nor a projection. More...

template<typename FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	UnpackRecvBuffers (const PackComponents &components, FabArray< FAB > &dest, const CommData &recv)
	De-serializes FAB data without any knowledge of a DTOS nor a projection. More...

template<typename FAB , typename DTOS , typename FabProj >
std::enable_if_t< IsBaseFab< FAB >::value >	LocalCopy (const ApplyDtosAndProjectionOnReciever< DTOS, FabProj > &packing, FabArray< FAB > &dest, const FabArray< FAB > &src, const FabArrayBase::CopyComTagsContainer &local_tags)
	Do local copies of FABs using DTOS and projection. More...

template<typename FAB , typename DTOS , typename FabProj >
std::enable_if_t< IsBaseFab< FAB >::value >	UnpackRecvBuffers (const ApplyDtosAndProjectionOnReciever< DTOS, FabProj > &packing, FabArray< FAB > &dest, const CommData &recv)
	Copy from received data in the buffer to destination FABs using DTOS and projection. More...

template<typename FAB , typename DataPacking , typename = std::enable_if_t<IsBaseFab<FAB>::value>, typename = std::enable_if_t<IsDataPacking<DataPacking, FAB>::value>>
AMREX_NODISCARD CommHandler	ParallelCopy_nowait (NoLocalCopy, FabArray< FAB > &dest, const FabArray< FAB > &src, const FabArrayBase::CommMetaData &cmd, const DataPacking &data_packing)

template<typename FAB , typename DataPacking , typename = std::enable_if_t<IsBaseFab<FAB>::value>, typename = std::enable_if_t<IsDataPacking<DataPacking, FAB>::value>>
AMREX_NODISCARD CommHandler	ParallelCopy_nowait (FabArray< FAB > &dest, const FabArray< FAB > &src, const FabArrayBase::CommMetaData &cmd, const DataPacking &data_packing)

template<typename FAB , typename DataPacking >
std::enable_if_t< IsBaseFab< FAB >) &&IsDataPacking< DataPacking, FAB >)>	ParallelCopy_finish (FabArray< FAB > &dest, CommHandler handler, const FabArrayBase::CommMetaData &cmd, const DataPacking &data_packing)

template<typename FAB , typename DataPacking >
std::enable_if_t< IsBaseFab< FAB >) &&IsDataPacking< DataPacking, FAB >)>	ParallelCopy_finish (DoLocalCopy, FabArray< FAB > &dest, const FabArray< FAB > &src, CommHandler handler, const FabArrayBase::CommMetaData &cmd, const DataPacking &data_packing)

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	ParallelCopy (FabArray< FAB > &dest, const FabArray< FAB > &src, const FabArrayBase::CommMetaData &cmd, SrcComp srccomp, DestComp destcomp, NumComps numcomp, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{})
	Call ParallelCopy_nowait followed by ParallelCopy_finish, strong typed version. More...

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	ParallelCopy (FabArray< FAB > &dest, const FabArray< FAB > &src, const FabArrayBase::CommMetaData &cmd, int srccomp, int destcomp, int numcomp, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{})
	Call ParallelCopy_nowait followed by ParallelCopy_finish. More...

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsIndexMapping< DTOS >) &&IsFabProjection< Proj, FAB >), MultiBlockCommMetaData >	ParallelCopy (FabArray< FAB > &dest, const Box &destbox, const FabArray< FAB > &src, SrcComp srccomp, DestComp destcomp, NumComps numcomp, const IntVect &ngrow, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{})
	Call ParallelCopy_nowait followed by ParallelCopy_finish, strong typed version. More...

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsIndexMapping< DTOS >) &&IsFabProjection< Proj, FAB >), MultiBlockCommMetaData >	ParallelCopy (FabArray< FAB > &dest, const Box &destbox, const FabArray< FAB > &src, int srccomp, int destcomp, int numcomp, const IntVect &ngrow, DTOS const &dtos=DTOS{}, Proj const &proj=Proj{})
	Call ParallelCopy_nowait followed by ParallelCopy_finish. More...

template<class FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	Rotate90 (FabArray< FAB > &mf, int scomp, int ncomp, IntVect const &nghost, Box const &domain)

template<class FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	Rotate90 (FabArray< FAB > &mf, Box const &domain)

template<class FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	Rotate180 (FabArray< FAB > &mf, int scomp, int ncomp, IntVect const &nghost, Box const &domain)

template<class FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	Rotate180 (FabArray< FAB > &mf, Box const &domain)

template<class FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	FillPolar (FabArray< FAB > &mf, int scomp, int ncomp, IntVect const &nghost, Box const &domain)

template<class FAB >
std::enable_if_t< IsBaseFab< FAB >::value >	FillPolar (FabArray< FAB > &mf, Box const &domain)

template<typename FAB , typename DTOS , typename Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >), CommHandler >	FillBoundary_nowait (FabArray< FAB > &mf, const FabArrayBase::CommMetaData &cmd, int scomp, int ncomp, DTOS const &dtos, Proj const &proj=Proj{})
	Start communication to fill boundary. More...

template<typename FAB , typename DTOS , typename Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	FillBoundary_finish (CommHandler handler, FabArray< FAB > &mf, const FabArrayBase::CommMetaData &cmd, int scomp, int ncomp, DTOS const &dtos, Proj const &proj=Proj{})
	Finish communication started by FillBoundary_nowait. More...

template<typename FAB , typename DTOS , typename Proj = Identity>
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >) &&IsFabProjection< Proj, FAB >)>	FillBoundary (FabArray< FAB > &mf, const FabArrayBase::CommMetaData &cmd, int scomp, int ncomp, DTOS const &dtos, Proj const &proj=Proj{})
	Fill ghost cells for FabArray/MultiFab. More...

template<typename FAB , typename DTOS >
std::enable_if_t< IsBaseFab< FAB >) &&IsCallableR< Dim3, DTOS, Dim3 >), FabArrayBase::CommMetaData >	makeFillBoundaryMetaData (FabArray< FAB > &mf, IntVect const &nghost, Geometry const &geom, DTOS const &dtos)
	Make metadata for FillBoundary. More...

Variables
static constexpr Identity	identity {}

template<int I, int J>
static constexpr SwapComponents< I, J >	swap_indices {}

static constexpr struct amrex::NonLocalBC::NoLocalCopy	no_local_copy

static constexpr struct amrex::NonLocalBC::DoLocalCopy	do_local_copy

Typedef Documentation

◆ DynamicSwapComponents

using amrex::NonLocalBC::DynamicSwapComponents = typedef SwapComponents<-1, -1>

◆ Inverse_t

template<typename T , typename... Args>

using amrex::NonLocalBC::Inverse_t = typedef decltype(std::declval<T>().Inverse(std::declval<Args>()...))

Return type of an InverseImage class member function.

◆ LocalCopy_t

template<typename... Args>

using amrex::NonLocalBC::LocalCopy_t = typedef decltype(LocalCopy(std::declval<Args>()...))

◆ PackSendBuffers_t

template<typename... Args>

using amrex::NonLocalBC::PackSendBuffers_t = typedef decltype(PackSendBuffers(std::declval<Args>()...))

◆ PrepareRecvBuffers_t

template<typename... Args>

using amrex::NonLocalBC::PrepareRecvBuffers_t = typedef decltype(PrepareRecvBuffers(std::declval<Args>()...))

◆ PrepareSendBuffers_t

template<typename... Args>

using amrex::NonLocalBC::PrepareSendBuffers_t = typedef decltype(PrepareSendBuffers(std::declval<Args>()...))

◆ UnpackRecvBuffers_t

template<typename... Args>

using amrex::NonLocalBC::UnpackRecvBuffers_t = typedef decltype(UnpackRecvBuffers(std::declval<Args>()...))

Function Documentation

◆ Apply()

template<typename DTOS >

std::enable_if_t<IsCallableR<Dim3, DTOS, Dim3>::value, IntVect> amrex::NonLocalBC::Apply	(	DTOS const &	dtos,
		const IntVect &	iv
	)

Applies the Dim3 to Dim3 mapping onto IntVects.

This is used to map indices from the dest index space into the source index space.

IntVect is being embedded in Dim3 by trailing zeros. Dim3 is being projected to IntVect by dopping z for AMREX_SPACEDIM = 2 or z and y components for AMREX_SPACEDIM = 1.

Parameters

[in] iv The IntVect that lives in the destination index space.

Returns: Returns IntVect{dtos(Dim3{iv})}

◆ ApplyInverse()

template<typename DTOS >

std::enable_if_t<HasInverseMemFn<DTOS>::value, IntVect> amrex::NonLocalBC::ApplyInverse	(	DTOS const &	dtos,
		const IntVect &	iv
	)

Applies the Dim3 to Dim3 invserse mapping onto IntVects.

This is used to map indices from the src index space into the dest index space.

IntVect is being embedded in Dim3 by trailing zeros. Dim3 is being projected to IntVect by dopping z for AMREX_SPACEDIM = 2 or z and y components for AMREX_SPACEDIM = 1.

Parameters

[in] iv The IntVect that lives in the src index space.

Returns: Returns IntVect{dtos.Inverse(Dim3{iv})}

◆ FillBoundary()

template<typename FAB , typename DTOS , typename Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3,DTOS,Dim3>) && IsFabProjection<Proj,FAB>)> amrex::NonLocalBC::FillBoundary	(	FabArray< FAB > &	mf,
		const FabArrayBase::CommMetaData &	cmd,
		int	scomp,
		int	ncomp,
		DTOS const &	dtos,
		Proj const &	proj = `Proj{}`
	)

Fill ghost cells for FabArray/MultiFab.

This fills ghost cells of a FabArray. This function is supposed to be used together with makeFillBoundaryMetaData as follows.

    auto cmd = makeFillBoundaryMetaData(mf, mf.nGrowVect, geom, dtos);
    // The metadata cmd can be cached and reused on a MultiFab/FabArray with
    // the same BoxArray and DistributionMapping.
    FillBoundary(mf, cmd, scomp, ncomp, dtos, proj);

The FillBoundary capability here is more flexible than FabArray's FillBoundary member functions, which only fill ghost cells inside the domain and ghost cells at periodic boundaries. The FillBoundary here can be used to fill non-local domain boundaries (e.g., in spherical and cylindrical coordinates) given appropriate index and component mappings.

Template Parameters

FAB	MultiFab/FabArray type
DTOS	Index mapping from destination from source. See SphThetaPhiRIndexMapping for an example.
Proj	Component mapping from source to destination. See SphThetaPhiRComponentMapping for an example.

Parameters

mf	FabArray/MultiFab whose ghost cells need to be filled.
cmd	communication metadata.
scomp	starting component.
ncomp	number of components.
dtos	index mapping.
proj	component mapping.

◆ FillBoundary_finish()

template<typename FAB , typename DTOS , typename Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3,DTOS,Dim3>) && IsFabProjection<Proj,FAB>)> amrex::NonLocalBC::FillBoundary_finish	(	CommHandler	handler,
		FabArray< FAB > &	mf,
		const FabArrayBase::CommMetaData &	cmd,
		int	scomp,
		int	ncomp,
		DTOS const &	dtos,
		Proj const &	proj = `Proj{}`
	)

Finish communication started by FillBoundary_nowait.

This finishes the communication to fill ghost cells of a FabArray. This function is supposed to be used together with FillBoundary_nowait and makeFillBoundaryMetaData as follows.

    auto cmd = makeFillBoundaryMetaData(mf, mf.nGrowVect, geom, dtos);
    // The metadata cmd can be cached and reused on a MultiFab/FabArray with
    // the same BoxArray and DistributionMapping.
    auto handler = FillBoundary_nowait(mf, cmd, scomp, ncomp, dtos, proj);
    // Independent computation can be performed.
    FillBoundary_finish(std::move(handler), mf, cmd, scomp, ncomp, dtos, proj);

The FillBoundary capability here is more flexible than FabArray's FillBoundary member functions, which only fill ghost cells inside the domain and ghost cells at periodic boundaries. The FillBoundary here can be used to fill non-local domain boundaries (e.g., in spherical and cylindrical coordinates) given appropriate index and component mappings.

Template Parameters

FAB	MultiFab/FabArray type
DTOS	Index mapping from destination from source. See SphThetaPhiRIndexMapping for an example.
Proj	Component mapping from source to destination. See SphThetaPhiRComponentMapping for an example.

Parameters

handler	CommHandler returned by FillBoundary_nowait.
mf	FabArray/MultiFab whose ghost cells need to be filled.
cmd	communication metadata.
scomp	starting component.
ncomp	number of components.
dtos	index mapping
proj	component mapping

◆ FillBoundary_nowait()

template<typename FAB , typename DTOS , typename Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3,DTOS,Dim3>) && IsFabProjection<Proj,FAB>), CommHandler> amrex::NonLocalBC::FillBoundary_nowait	(	FabArray< FAB > &	mf,
		const FabArrayBase::CommMetaData &	cmd,
		int	scomp,
		int	ncomp,
		DTOS const &	dtos,
		Proj const &	proj = `Proj{}`
	)

Start communication to fill boundary.

This starts communication to fill ghost cells of a FabArray. This function is supposed to be used together with FillBoundary_finish and makeFillBoundaryMetaData as follows.

    auto cmd = makeFillBoundaryMetaData(mf, mf.nGrowVect, geom, dtos);
    // The metadata cmd can be cached and reused on a MultiFab/FabArray with
    // the same BoxArray and DistributionMapping.
    auto handler = FillBoundary_nowait(mf, cmd, scomp, ncomp, dtos, proj);
    // Independent computation can be performed.
    FillBoundary_finish(std::move(handler), mf, cmd, scomp, ncomp, dtos, proj);

The FillBoundary capability here is more flexible than FabArray's FillBoundary member functions, which only fill ghost cells inside the domain and ghost cells at periodic boundaries. The FillBoundary here can be used to fill non-local domain boundaries (e.g., in spherical and cylindrical coordinates) given appropriate index and component mappings.

Template Parameters

FAB	MultiFab/FabArray type
DTOS	Index mapping from destination from source. See SphThetaPhiRIndexMapping for an example.
Proj	Component mapping from source to destination. See SphThetaPhiRComponentMapping for an example.

Parameters

mf	FabArray/MultiFab whose ghost cells need to be filled.
cmd	communication metadata.
scomp	starting component.
ncomp	number of components.
dtos	index mapping.
proj	component mapping.

Returns: a CommHandler object needed for calling FillBoundary_finish.

◆ FillPolar() [1/2]

template<class FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::FillPolar	(	FabArray< FAB > &	mf,
		Box const &	domain
	)

◆ FillPolar() [2/2]

template<class FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::FillPolar	(	FabArray< FAB > &	mf,
		int	scomp,
		int	ncomp,
		IntVect const &	nghost,
		Box const &	domain
	)

◆ Image() [1/2]

template<typename DTOS >

std::enable_if_t<IsCallableR<Dim3, DTOS, Dim3>::value && !IsCallableR<IndexType, DTOS, IndexType>::value, Box> amrex::NonLocalBC::Image	(	DTOS const &	dtos,
		const Box &	box
	)

Applies the Dim3 to Dim3 mapping onto Boxes but does not change the index type.

This function assumes monotonicity of dtos in each component.

Parameters

[in] box The box that lives the destination index space

Returns: Returns the smallest Box in the source index space that contains images of Apply(dtos, box.smallEnd()) and Apply(dtos, box.bigEnd()).

◆ Image() [2/2]

template<typename DTOS >

std::enable_if_t<IsCallableR<Dim3, DTOS, Dim3>::value && IsCallableR<IndexType, DTOS, IndexType>::value, Box> amrex::NonLocalBC::Image	(	DTOS const &	dtos,
		const Box &	box
	)

Applies the Dim3 to Dim3 mapping onto Boxes and maps the index type.

This function assumes monotonicity of dtos in each component.

Parameters

[in] box The box that lives the destination index space

Returns: Returns the smallest Box in the source index space that contains images of Apply(dtos, box.smallEnd()) and Apply(dtos, box.bigEnd()).

◆ InverseImage() [1/2]

template<typename DTOS >

std::enable_if_t<HasInverseMemFn<DTOS>::value && !IsCallableR<IndexType, DTOS, IndexType>::value, Box> amrex::NonLocalBC::InverseImage	(	DTOS const &	dtos,
		const Box &	box
	)

Applies the inverse Dim3 to Dim3 mapping onto Boxes without changing the index type.

This function assumes monotonicity in each component.

Parameters

[in] box The box that lives the source index space

Returns: Returns the smallest Box in the destination index space that contains images of ApplyInverse(box.smallEnd()) and ApplyInverse(box.bigEnd()).

◆ InverseImage() [2/2]

template<typename DTOS >

std::enable_if_t<HasInverseMemFn<DTOS>::value && IsCallableR<IndexType, DTOS, IndexType>::value, Box> amrex::NonLocalBC::InverseImage	(	DTOS const &	dtos,
		const Box &	box
	)

Applies the inverse Dim3 to Dim3 mapping onto Boxes.

This function assumes monotonicity in each component.

Parameters

[in] box The box that lives the source index space

Returns: Returns the smallest Box in the destination index space that contains images of ApplyInverse(box.smallEnd()) and ApplyInverse(box.bigEnd()).

◆ local_copy_cpu()

template<class FAB , class DTOS = Identity, class Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3, DTOS, Dim3>) && IsFabProjection<Proj, FAB>)> amrex::NonLocalBC::local_copy_cpu	(	FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		int	dcomp,
		int	scomp,
		int	ncomp,
		FabArrayBase::CopyComTagsContainer const &	local_tags,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

noexcept

◆ local_copy_gpu()

template<class FAB , class DTOS = Identity, class Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3, DTOS, Dim3>) && IsFabProjection<Proj, FAB>)> amrex::NonLocalBC::local_copy_gpu	(	FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		int	dcomp,
		int	scomp,
		int	ncomp,
		FabArrayBase::CopyComTagsContainer const &	local_tags,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

noexcept

◆ LocalCopy() [1/2]

template<typename FAB , typename DTOS , typename FabProj >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::LocalCopy	(	const ApplyDtosAndProjectionOnReciever< DTOS, FabProj > &	packing,
		FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		const FabArrayBase::CopyComTagsContainer &	local_tags
	)

Do local copies of FABs using DTOS and projection.

◆ LocalCopy() [2/2]

template<typename FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::LocalCopy	(	const PackComponents &	components,
		FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		const FabArrayBase::CopyComTagsContainer &	local_tags
	)

Dispatch local copies to the default behaviour that knows no DTOS nor projection.

◆ makeFillBoundaryMetaData()

template<typename FAB , typename DTOS >

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3,DTOS,Dim3>), FabArrayBase::CommMetaData> amrex::NonLocalBC::makeFillBoundaryMetaData	(	FabArray< FAB > &	mf,
		IntVect const &	nghost,
		Geometry const &	geom,
		DTOS const &	dtos
	)

Make metadata for FillBoundary.

Template Parameters

FAB	MultiFab/FabArray type
DTOS	Index mapping from destination from source. See SphThetaPhiRIndexMapping for an example.

Parameters

mf	FabArray/MultiFab whose ghost cells need to be filled.
nghost	number of ghost cells to be filled.
geom	a Geometry object that contains the domain information.
dtos	index mapping.

Returns: communication metadata

◆ PackSendBuffers()

template<typename FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::PackSendBuffers	(	const PackComponents &	components,
		const FabArray< FAB > &	src,
		CommData &	send
	)

Serializes FAB data without any knowledge of a DTOS nor a projection.

◆ ParallelCopy() [1/5]

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsIndexMapping<DTOS>) && IsFabProjection<Proj, FAB>),MultiBlockCommMetaData> amrex::NonLocalBC::ParallelCopy	(	FabArray< FAB > &	dest,
		const Box &	destbox,
		const FabArray< FAB > &	src,
		int	srccomp,
		int	destcomp,
		int	numcomp,
		const IntVect &	ngrow,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

Call ParallelCopy_nowait followed by ParallelCopy_finish.

This function constructs a new MultiCommMetaData from the given DTOS, destbox and ngrow.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	destbox	The index box in the destination space that will be filled by data from src. The source box that describes the dependencies will be computed by the specified DTOS.
[in]	src	The Multifab that is used to fill the send buffers.
[in]	srccomp	The first component in src that will be copied to dest.
[in]	destcomp	The first component in dest that will get written by src.
[in]	ncomp	The number of successive components that will be copied.
[in]	ngrow	The amount of ghost cells that will be taking into consideration. Note, even if destbox contains indices outside the domain we need to specify an appropriate ngrow that covers the amount of ghost cells that we want to copy.
[in]	dtos	An index mapping that maps indices from destination space to source space and from source space to destination space.
[in]	proj	A transformation function that might change the data when it is being copied.

Returns: Returns the CommMetaData object that can be cached for future calls to ParallelCopy.

◆ ParallelCopy() [2/5]

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsIndexMapping<DTOS>) && IsFabProjection<Proj, FAB>),MultiBlockCommMetaData> amrex::NonLocalBC::ParallelCopy	(	FabArray< FAB > &	dest,
		const Box &	destbox,
		const FabArray< FAB > &	src,
		SrcComp	srccomp,
		DestComp	destcomp,
		NumComps	numcomp,
		const IntVect &	ngrow,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

Call ParallelCopy_nowait followed by ParallelCopy_finish, strong typed version.

This function constructs a new MultiCommMetaData from the given DTOS, destbox and ngrow.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	destbox	The index box in the destination space that will be filled by data from src. The source box that describes the dependencies will be computed by the specified DTOS.
[in]	src	The Multifab that is used to fill the send buffers.
[in]	srccomp	The first component in src that will be copied to dest.
[in]	destcomp	The first component in dest that will get written by src.
[in]	ncomp	The number of successive components that will be copied.
[in]	ngrow	The amount of ghost cells that will be taking into consideration. Note, even if destbox contains indices outside the domain we need to specify an appropriate ngrow that covers the amount of ghost cells that we want to copy.
[in]	dtos	An index mapping that maps indices from destination space to source space and from source space to destination space.
[in]	proj	A transformation function that might change the data when it is being copied.

Returns: Returns the CommMetaData object that can be cached for future calls to ParallelCopy.

◆ ParallelCopy() [3/5]

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3, DTOS, Dim3>) && IsFabProjection<Proj, FAB>)> amrex::NonLocalBC::ParallelCopy	(	FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		const FabArrayBase::CommMetaData &	cmd,
		int	srccomp,
		int	destcomp,
		int	numcomp,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

Call ParallelCopy_nowait followed by ParallelCopy_finish.

This function overload uses an already cached CommMetaData. This CommMetaData needs to be compatible with the specified DTOS and projection, otherwise undefined behaviour occurs.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	src	The Multifab that is used to fill the send buffers.
[in]	cmd	The communication meta data object holds spatial information about FAB boxes that need to be filled and copied from.
[in]	srccomp	The first component in src that will be copied to dest.
[in]	destcomp	The first component in dest that will get written by src.
[in]	ncomp	The number of successive components that will be copied.
[in]	dtos	An index mapping that maps indices from destination space to source space and from source space to destination space.
[in]	proj	A transformation function that might change the data when it is being copied.

Returns: Nothing.

◆ ParallelCopy() [4/5]

template<typename FAB , typename DTOS = Identity, typename Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3, DTOS, Dim3>) && IsFabProjection<Proj, FAB>)> amrex::NonLocalBC::ParallelCopy	(	FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		const FabArrayBase::CommMetaData &	cmd,
		SrcComp	srccomp,
		DestComp	destcomp,
		NumComps	numcomp,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

Call ParallelCopy_nowait followed by ParallelCopy_finish, strong typed version.

This function overload uses an already cached CommMetaData. This CommMetaData needs to be compatible with the specified DTOS and projection, otherwise undefined behaviour occurs.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	src	The Multifab that is used to fill the send buffers.
[in]	cmd	The communication meta data object holds spatial information about FAB boxes that need to be filled and copied from.
[in]	srccomp	The first component in src that will be copied to dest.
[in]	destcomp	The first component in dest that will get written by src.
[in]	ncomp	The number of successive components that will be copied.
[in]	dtos	An index mapping that maps indices from destination space to source space and from source space to destination space.
[in]	proj	A transformation function that might change the data when it is being copied.

Returns: Nothing.

◆ ParallelCopy() [5/5]

template MultiBlockCommMetaData amrex::NonLocalBC::ParallelCopy	(	FabArray< FArrayBox > &	dest,
		const Box &	destbox,
		const FabArray< FArrayBox > &	src,
		int	destcomp,
		int	srccomp,
		int	numcomp,
		const IntVect &	ngrow,
		MultiBlockIndexMapping const &	,
		Identity const &
	)

◆ ParallelCopy_finish() [1/2]

template<typename FAB , typename DataPacking >

std::enable_if_t<IsBaseFab<FAB>) && IsDataPacking<DataPacking, FAB>)> amrex::NonLocalBC::ParallelCopy_finish	(	DoLocalCopy	,
		FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		CommHandler	handler,
		const FabArrayBase::CommMetaData &	cmd,
		const DataPacking &	data_packing
	)

Blockingly wait for all communication to be done and fill the local FABs with received data.

This function overload performs local copies, i.e. from this MPI process to itself. It will block the current thread until all MPI recv and send requests are done and calls the DataPacking object to unpack the received buffers.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	src	The Multifab that is used to get the data for the local copies from.
[in]	handler	This object holds all data buffers that need to be kept alive as long as the data transaction is not done.
[in]	cmd	The communication meta data object holds spatial information about FAB boxes that need to be filled.
[in]	data_packing	A CPO that controls behaviour of unpacking the received data.

Returns: Nothing.

◆ ParallelCopy_finish() [2/2]

template<typename FAB , typename DataPacking >

std::enable_if_t<IsBaseFab<FAB>) && IsDataPacking<DataPacking, FAB>)> amrex::NonLocalBC::ParallelCopy_finish	(	FabArray< FAB > &	dest,
		CommHandler	handler,
		const FabArrayBase::CommMetaData &	cmd,
		const DataPacking &	data_packing
	)

Blockingly wait for all communication to be done and fill the local FABs with received data.

This function overload performs no local copies, i.e. from this MPI process to itself. It will block the current thread until all MPI recv and send requests are done and calls the DataPacking object to unpack the received buffers.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	handler	This object holds all data buffers that need to be kept alive as long as the data transaction is not done.
[in]	cmd	The communication meta data object holds spatial information about FAB boxes that need to be filled.
[in]	data_packing	A CPO that controls behaviour of unpacking the received buffer to the destination FabArray.

Returns: Nothing.

◆ ParallelCopy_nowait() [1/2]

template<typename FAB , typename DataPacking , typename = std::enable_if_t<IsBaseFab<FAB>::value>, typename = std::enable_if_t<IsDataPacking<DataPacking, FAB>::value>>

AMREX_NODISCARD CommHandler amrex::NonLocalBC::ParallelCopy_nowait	(	FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		const FabArrayBase::CommMetaData &	cmd,
		const DataPacking &	data_packing
	)

Initiate recv and send calls for MPI and return after doing local work.

DataPacking is a customization point object to control the behaviour of packing and unpacking send or recv data buffers. It is used to perform interpolation or data transformations on either sender or receiver side.

This function performs a data packing on sender side and we expect a call to Parallel_finish that performs data unpacking on the receiver side.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	src	The Multifab that is used to fill the send buffers.
[in]	cmd	The communication meta data object holds spatial information about FAB boxes that need to be filled and copied from.
[in]	data_packing	A CPO that controls behaviour of preparing buffers and packing the source data into the send buffers.

Returns: Returns a CommHandler object that owns context and memory buffers for the whole life time of the MPI transaction.

◆ ParallelCopy_nowait() [2/2]

template<typename FAB , typename DataPacking , typename = std::enable_if_t<IsBaseFab<FAB>::value>, typename = std::enable_if_t<IsDataPacking<DataPacking, FAB>::value>>

AMREX_NODISCARD CommHandler amrex::NonLocalBC::ParallelCopy_nowait	(	NoLocalCopy	,
		FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		const FabArrayBase::CommMetaData &	cmd,
		const DataPacking &	data_packing
	)

Initiate recv and send calls for MPI and immediately return without doing any work.

DataPacking is a customization point object to control the behaviour of packing and unpacking send or recv data buffers. It is used to perform interpolation or data transformations on either sender or receiver side.

This function performs a data packing on sender side and we expect a call to Parallel_finish that performs data unpacking on the receiver side.

Parameters

[out]	dest	The Multifab that is going to be filled with received data.
[in]	src	The Multifab that is used to fill the send buffers.
[in]	cmd	The communication meta data object holds spatial information about FAB boxes that need to be filled and copied from.
[in]	data_packing	A CPO that controls behaviour of preparing buffers and packing the source data into the send buffers.

Returns: Returns a CommHandler object that owns context and memory buffers for the whole life time of the MPI transaction.

◆ PostRecvs()

void amrex::NonLocalBC::PostRecvs	(	CommData &	recv,
		int	mpi_tag
	)

Initiate all recvieves with MPI_Irecv calls associated with tag mpi_tag.

◆ PostSends()

void amrex::NonLocalBC::PostSends	(	CommData &	send,
		int	mpi_tag
	)

Initiate all sends with MPI_Isend calls associated with tag mpi_tag.

◆ PrepareCommBuffers()

void amrex::NonLocalBC::PrepareCommBuffers	(	CommData &	comm,
		const FabArrayBase::MapOfCopyComTagContainers &	cctc,
		int	n_components,
		std::size_t	object_size,
		std::size_t	align
	)

Fill all class member variables of comm but the request and the stats vector.

◆ PrepareRecvBuffers()

template<typename FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::PrepareRecvBuffers	(	const PackComponents &	components,
		FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		CommData &	comm,
		const FabArrayBase::MapOfCopyComTagContainers &	cctc
	)

Calls PrepareComBuffers.

◆ PrepareSendBuffers()

template<typename FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::PrepareSendBuffers	(	const PackComponents &	components,
		FabArray< FAB > &	dest,
		const FabArray< FAB > &	src,
		CommData &	comm,
		const FabArrayBase::MapOfCopyComTagContainers &	cctc
	)

Calls PrepareComBuffers.

◆ Rotate180() [1/2]

template<class FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::Rotate180	(	FabArray< FAB > &	mf,
		Box const &	domain
	)

◆ Rotate180() [2/2]

template<class FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::Rotate180	(	FabArray< FAB > &	mf,
		int	scomp,
		int	ncomp,
		IntVect const &	nghost,
		Box const &	domain
	)

◆ Rotate90() [1/2]

template<class FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::Rotate90	(	FabArray< FAB > &	mf,
		Box const &	domain
	)

◆ Rotate90() [2/2]

template<class FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::Rotate90	(	FabArray< FAB > &	mf,
		int	scomp,
		int	ncomp,
		IntVect const &	nghost,
		Box const &	domain
	)

◆ unpack_recv_buffer_cpu()

template<class FAB , class DTOS = Identity, class Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3, DTOS, Dim3>) && IsFabProjection<Proj, FAB>)> amrex::NonLocalBC::unpack_recv_buffer_cpu	(	FabArray< FAB > &	mf,
		int	dcomp,
		int	ncomp,
		Vector< char * > const &	recv_data,
		Vector< std::size_t > const &	recv_size,
		Vector< FabArrayBase::CopyComTagsContainer const * > const &	recv_cctc,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

noexcept

◆ unpack_recv_buffer_gpu()

template<class FAB , class DTOS = Identity, class Proj = Identity>

std::enable_if_t<IsBaseFab<FAB>) && IsCallableR<Dim3, DTOS, Dim3>) && IsFabProjection<Proj, FAB>)> amrex::NonLocalBC::unpack_recv_buffer_gpu	(	FabArray< FAB > &	mf,
		int	scomp,
		int	ncomp,
		Vector< char * > const &	recv_data,
		Vector< std::size_t > const &	recv_size,
		Vector< FabArrayBase::CopyComTagsContainer const * > const &	recv_cctc,
		DTOS const &	dtos = `DTOS{}`,
		Proj const &	proj = `Proj{}`
	)

◆ UnpackRecvBuffers() [1/2]

template<typename FAB , typename DTOS , typename FabProj >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::UnpackRecvBuffers	(	const ApplyDtosAndProjectionOnReciever< DTOS, FabProj > &	packing,
		FabArray< FAB > &	dest,
		const CommData &	recv
	)

Copy from received data in the buffer to destination FABs using DTOS and projection.

◆ UnpackRecvBuffers() [2/2]

template<typename FAB >

std::enable_if_t<IsBaseFab<FAB>::value> amrex::NonLocalBC::UnpackRecvBuffers	(	const PackComponents &	components,
		FabArray< FAB > &	dest,
		const CommData &	recv
	)

De-serializes FAB data without any knowledge of a DTOS nor a projection.

Variable Documentation

◆ do_local_copy

constexpr struct amrex::NonLocalBC::DoLocalCopy amrex::NonLocalBC::do_local_copy

static

◆ identity

constexpr Identity amrex::NonLocalBC::identity {}

staticconstexpr

◆ no_local_copy

constexpr struct amrex::NonLocalBC::NoLocalCopy amrex::NonLocalBC::no_local_copy

static

◆ swap_indices

template<int I, int J>

constexpr SwapComponents<I, J> amrex::NonLocalBC::swap_indices {}

staticconstexpr

Namespaces

Classes

Typedefs

Functions

Variables

Typedef Documentation

◆ DynamicSwapComponents

◆ Inverse_t

◆ LocalCopy_t

◆ PackSendBuffers_t

◆ PrepareRecvBuffers_t

◆ PrepareSendBuffers_t

◆ UnpackRecvBuffers_t

Function Documentation

◆ Apply()

◆ ApplyInverse()

◆ FillBoundary()

◆ FillBoundary_finish()

◆ FillBoundary_nowait()

◆ FillPolar() [1/2]

◆ FillPolar() [2/2]

◆ Image() [1/2]

◆ Image() [2/2]

◆ InverseImage() [1/2]

◆ InverseImage() [2/2]

◆ local_copy_cpu()

◆ local_copy_gpu()

◆ LocalCopy() [1/2]

◆ LocalCopy() [2/2]

◆ makeFillBoundaryMetaData()

◆ PackSendBuffers()

◆ ParallelCopy() [1/5]

◆ ParallelCopy() [2/5]

◆ ParallelCopy() [3/5]

◆ ParallelCopy() [4/5]

◆ ParallelCopy() [5/5]

◆ ParallelCopy_finish() [1/2]

◆ ParallelCopy_finish() [2/2]

◆ ParallelCopy_nowait() [1/2]

◆ ParallelCopy_nowait() [2/2]

◆ PostRecvs()

◆ PostSends()

◆ PrepareCommBuffers()

◆ PrepareRecvBuffers()

◆ PrepareSendBuffers()

◆ Rotate180() [1/2]

◆ Rotate180() [2/2]

◆ Rotate90() [1/2]

◆ Rotate90() [2/2]

◆ unpack_recv_buffer_cpu()

◆ unpack_recv_buffer_gpu()

◆ UnpackRecvBuffers() [1/2]

◆ UnpackRecvBuffers() [2/2]

Variable Documentation

◆ do_local_copy

◆ identity

◆ no_local_copy

◆ swap_indices