par Namespace Reference

Collection of Generic Parallel Functions: Sorting, Partitioning, Searching,... More...

Classes
class	par::Mpi_datatype
	An abstract class used for communicating messages using user-defined datatypes. The user must implement the static member function "value()" that returns the MPI_Datatype corresponding to this user-defined datatype. More...
class	par::Mpi_datatype< bool >
	A template specialization of the abstract class Mpi_datatype. This can be used for communicating messages of type "bool". More...
class	par::Mpi_datatype< MeshRecord >
class	par::Mpi_datatype< ot::FineTouchedDummyStatus >
class	par::Mpi_datatype< ot::FineTouchedStatus >
class	par::Mpi_datatype< ot::NodeAndValues< T, ARR_LEN > >
	A template specialization of the abstract class "Mpi_datatype" for communicating messages of type "ot::NodeAndValues<T, unsigned int>". More...
class	par::Mpi_datatype< ot::TreeNode >
	A template specialization of the abstract class "Mpi_datatype" for communicating messages of type "ot::TreeNode". More...
class	par::Mpi_datatype< std::complex< T > >
class	par::Mpi_datatype< Vertex >
Functions
template<typename T>
void	bitonicSort (std::vector< T > &in, MPI_Comm comm)
	An implementation of parallel bitonic sort that does not expect the number of processors to be a power of 2. In fact, the number of processors can even be odd. Moreover, we do not even expect the length of the vector (neither locally nor globally) to be a power of 2 or even. Moreover, each processor can call this with a different number of elements. However, we do expect that 'in' atleast has 1 element on each processor. This recursively calls the function bitonicSort_binary, followed by a special parallel merge.
template<typename T>
void	bitonicSort_binary (std::vector< T > &in, MPI_Comm comm)
	An implementation of parallel bitonic sort that expects the number of processors to be a power of 2. However, unlike most implementations, we do not expect the length of the vector (neither locally nor globally) to be a power of 2 or even. Moreover, each processor can call this with a different number of elements. However, we do expect that 'in' atleast has 1 element on each processor.
template<typename T>
int	concatenate (std::vector< T > &listA, std::vector< T > &listB, MPI_Comm comm)
	A parallel concatenation function. listB is appended (globally) to listA and the result is stored in listA. An useful application of this function is when listA and listB are individually sorted (globally) and the smallest element in listB is greater than the largest element in listA and we want to create a merged list that is sorted.
template<typename T>
unsigned int	defaultWeight (const T *a)
template<typename T>
int	maxLowerBound (const std::vector< T > &keys, const std::vector< T > &searchList, std::vector< T > &results, MPI_Comm comm)
	A parallel search function.
template<typename T>
void	MergeLists (std::vector< T > &listA, std::vector< T > &listB, int KEEP_WHAT)
	Merges lists A, and B, retaining either the low or the High in list A.
template<typename T>
void	MergeSplit (std::vector< T > &local_list, int which_keys, int partner, MPI_Comm comm)
	The main operation in the parallel bitonic sort algorithm. This implements the compare-split operation.
template<typename T>
int	Mpi_Allgather (T *sendbuf, T *recvbuf, int count, MPI_Comm comm)
template<typename T>
int	Mpi_Allgatherv (T *sendbuf, int sendcount, T *recvbuf, int *recvcounts, int *displs, MPI_Comm comm)
template<typename T>
int	Mpi_Allreduce (T *sendbuf, T *recvbuf, int count, MPI_Op op, MPI_Comm comm)
template<typename T>
int	Mpi_Alltoall (T *sendbuf, T *recvbuf, int count, MPI_Comm comm)
template<typename T>
int	Mpi_Alltoallv_dense (T *sendbuf, int *sendcnts, int *sdispls, T *recvbuf, int *recvcnts, int *rdispls, MPI_Comm comm)
template<typename T>
int	Mpi_Alltoallv_sparse (T *sendbuf, int *sendcnts, int *sdispls, T *recvbuf, int *recvcnts, int *rdispls, MPI_Comm comm)
template<typename T>
int	Mpi_Bcast (T *buffer, int count, int root, MPI_Comm comm)
template<typename T>
int	Mpi_Gather (T *sendBuffer, T *recvBuffer, int count, int root, MPI_Comm comm)
template<typename T>
int	Mpi_Irecv (T *buf, int count, int source, int tag, MPI_Comm comm, MPI_Request *request)
template<typename T>
int	Mpi_Issend (T *buf, int count, int dest, int tag, MPI_Comm comm, MPI_Request *request)
template<typename T>
int	Mpi_Recv (T *buf, int count, int source, int tag, MPI_Comm comm, MPI_Status *status)
template<typename T>
int	Mpi_Reduce (T *sendbuf, T *recvbuf, int count, MPI_Op op, int root, MPI_Comm comm)
template<typename T>
int	Mpi_Scan (T *sendbuf, T *recvbuf, int count, MPI_Op op, MPI_Comm comm)
template<typename T, typename S>
int	Mpi_Sendrecv (T *sendBuf, int sendCount, int dest, int sendTag, S *recvBuf, int recvCount, int source, int recvTag, MPI_Comm comm, MPI_Status *status)
template<typename T>
void	Par_bitonic_merge_incr (std::vector< T > &local_list, int proc_set_size, MPI_Comm comm)
template<typename T>
void	Par_bitonic_sort_decr (std::vector< T > &local_list, int proc_set_size, MPI_Comm comm)
template<typename T>
void	Par_bitonic_sort_incr (std::vector< T > &local_list, int proc_set_size, MPI_Comm comm)
template<typename T>
int	partitionW (std::vector< T > &vec, unsigned int(*getWeight)(const T *), MPI_Comm comm)
	A parallel weighted partitioning function. In our implementation, we do not pose any restriction on the input or the number of processors. This function can be used with an odd number of processors as well. Some processors can pass an empty vector as input. The relative ordering of the elements is preserved.
template<typename T>
int	removeDuplicates (std::vector< T > &nodes, bool isSorted, MPI_Comm comm)
	Removes duplicates in parallel. If the input is not sorted, sample sort will be called within the function to sort the vector and then duplicates will be removed.
template<typename T>
int	sampleSort (std::vector< T > &in, std::vector< T > &out, MPI_Comm comm)
	A parallel sample sort implementation. In our implementation, we do not pose any restriction on the input or the number of processors. This function can be used with an odd number of processors as well. Some processors can pass an empty vector as input. If the total number of elements in the vector (globally) is fewer than 10*p^2, where p is the number of processors, then we will use bitonic sort instead of sample sort to sort the vector. We use a paralle bitonic sort to sort the samples in the sample sort algorithm. Hence, the complexity of the algorithm is O(n/p log n/p) + O(p log p). Here, n is the global length of the vector and p is the number of processors.
template<typename T>
int	scatterValues (std::vector< T > &in, std::vector< T > &out, DendroIntL outSz, MPI_Comm comm)
	Re-distributes a STL vector, preserving the relative ordering of the elements.
int	splitComm2way (const bool *isEmptyList, MPI_Comm *new_comm, MPI_Comm comm)
	Splits a communication group into two depending on the values in isEmptyList. Both the groups are sorted in the ascending order of their ranks in the old comm. All processors must call this function with the same 'isEmptyList' array.
int	splitComm2way (bool iAmEmpty, MPI_Comm *new_comm, MPI_Comm comm)
	Splits a communication group into two, one containing processors that passed a value of 'false' for the parameter 'iAmEmpty' and the another containing processors that passed a value of 'true' for the parameter. Both the groups are sorted in the ascending order of their ranks in the old comm.
unsigned int	splitCommBinary (MPI_Comm orig_comm, MPI_Comm *new_comm)
	Splits a communication group into two, the first having a power of 2 number of processors and the other having the remainder. The first group is sorted in the ascending order of their ranks in the old comm and the second group is sorted in the descending order of their ranks in the old comm.
unsigned int	splitCommBinaryNoFlip (MPI_Comm orig_comm, MPI_Comm *new_comm)
	Splits a communication group into two, the first having a power of 2 number of processors and the other having the remainder. Both the groups are sorted in the ascending order of their ranks in the old comm.
int	splitCommUsingSplittingRank (int splittingRank, MPI_Comm *new_comm, MPI_Comm comm)

---

Detailed Description

Collection of Generic Parallel Functions: Sorting, Partitioning, Searching,...

Author:

Rahul Sampath

Hari Sundar

---

Function Documentation

template<typename T> void par::bitonicSort (  std::vector< T > &  in, MPI_Comm  comm )

An implementation of parallel bitonic sort that does not expect the number of processors to be a power of 2. In fact, the number of processors can even be odd. Moreover, we do not even expect the length of the vector (neither locally nor globally) to be a power of 2 or even. Moreover, each processor can call this with a different number of elements. However, we do expect that 'in' atleast has 1 element on each processor. This recursively calls the function bitonicSort_binary, followed by a special parallel merge. Parameters: in  the vector to be sorted Author: Hari Sundar See also: bitonicSort_binary Definition at line 1623 of file parUtils.txx. References Par_bitonic_merge_incr(), sort(), splitCommBinary(), and splitCommBinaryNoFlip().

template<typename T> void par::bitonicSort_binary (  std::vector< T > &  in, MPI_Comm  comm )

An implementation of parallel bitonic sort that expects the number of processors to be a power of 2. However, unlike most implementations, we do not expect the length of the vector (neither locally nor globally) to be a power of 2 or even. Moreover, each processor can call this with a different number of elements. However, we do expect that 'in' atleast has 1 element on each processor. Parameters: in  the vector to be sorted Author: Hari Sundar Definition at line 1593 of file parUtils.txx.

template<typename T> int par::concatenate (  std::vector< T > &  listA, std::vector< T > &  listB, MPI_Comm  comm )

A parallel concatenation function. listB is appended (globally) to listA and the result is stored in listA. An useful application of this function is when listA and listB are individually sorted (globally) and the smallest element in listB is greater than the largest element in listA and we want to create a merged list that is sorted. Parameters: listA  a distributed vector, the result is stored in listA listB  another distributed vector that is appended to listA listA must not be empty on any of the calling processors. listB can be empty on some of the calling processors. listB will be cleared within the function. comm  the communicator Author: Rahul Sampath Definition at line 547 of file parUtils.txx. References DendroIntL, PROF_PAR_CONCAT_BEGIN, and PROF_PAR_CONCAT_END.

template<typename T> unsigned int par::defaultWeight (  const T *  a  )

Definition at line 393 of file parUtils.txx.

template<typename T> int par::maxLowerBound (  const std::vector< T > &  keys, const std::vector< T > &  searchList, std::vector< T > &  results, MPI_Comm  comm )

A parallel search function. Author: Rahul Sampath Parameters: keys  locally sorted unique list of keys searchList  globally sorted unique list. No processor must call this function with an empty list. results  maximum lower bound in searchList for the corresponding key comm  MPI communicator Returns: int errorcode Definition at line 764 of file parUtils.txx. References PROF_SEARCH_BEGIN, and PROF_SEARCH_END.

template<typename T> void par::MergeLists (  std::vector< T > &  listA, std::vector< T > &  listB, int  KEEP_WHAT )

Merges lists A, and B, retaining either the low or the High in list A. Author: Hari Sundar Parameters: listA  Input list, and where the output is stored. listB  Second input list. KEEP_WHAT  determines whether to retain the High or the low values from A and B. One of KEEP_HIGH or KEEP_LOW. Merging the two lists when their sizes are not the same is a bit involved. The major condition that needs to be used is that all elements that are less than max(min(A), min(B)) are retained by the KEEP_LOW processor, and similarly all elements that are larger larger than min(max(A), max(B)) are retained by the KEEP_HIGH processor. The reason for this is that, on the Keep_Low side, max(min(A), min(B)) > min(A) > max(A-) and similarly on the Keep_high side, min(max(A), max(B)) < max(A) < min(A+) which guarantees that the merged lists remain bitonic. Definition at line 1673 of file parUtils.txx.

template<typename T> void par::MergeSplit (  std::vector< T > &  local_list, int  which_keys, int  partner, MPI_Comm  comm )

The main operation in the parallel bitonic sort algorithm. This implements the compare-split operation. Author: Hari Sundar Parameters: which_keys  is one of KEEP_HIGH or KEEP_LOW partner  is the processor with which to Merge and Split. local_list  the input vector comm  the communicator Definition at line 1476 of file parUtils.txx.

template<typename T> int par::Mpi_Allgather (  T *  sendbuf, T *  recvbuf, int  count, MPI_Comm  comm )

Author: Rahul S. Sampath Definition at line 196 of file parUtils.txx. References PROF_PAR_ALLGATHER_BEGIN, and PROF_PAR_ALLGATHER_END.

template<typename T> int par::Mpi_Allgatherv (  T *  sendbuf, int  sendcount, T *  recvbuf, int *  recvcounts, int *  displs, MPI_Comm  comm )

Author: Rahul S. Sampath Definition at line 147 of file parUtils.txx. References PROF_PAR_ALLGATHERV_BEGIN, and PROF_PAR_ALLGATHERV_END.

template<typename T> int par::Mpi_Allreduce (  T *  sendbuf, T *  recvbuf, int  count, MPI_Op  op, MPI_Comm  comm )  [inline]

Author: Rahul S. Sampath Definition at line 85 of file parUtils.txx. References PROF_PAR_ALLREDUCE_BEGIN, and PROF_PAR_ALLREDUCE_END.

template<typename T> int par::Mpi_Alltoall (  T *  sendbuf, T *  recvbuf, int  count, MPI_Comm  comm )  [inline]

Author: Rahul S. Sampath Definition at line 97 of file parUtils.txx. References PROF_PAR_ALL2ALL_BEGIN, and PROF_PAR_ALL2ALL_END. Referenced by ot::flagNodesType2().

template<typename T> int par::Mpi_Alltoallv_dense (  T *  sendbuf, int *  sendcnts, int *  sdispls, T *  recvbuf, int *  recvcnts, int *  rdispls, MPI_Comm  comm )

Author: Rahul S. Sampath Definition at line 318 of file parUtils.txx. References PROF_PAR_ALL2ALLV_DENSE_BEGIN, and PROF_PAR_ALL2ALLV_DENSE_END.

template<typename T> int par::Mpi_Alltoallv_sparse (  T *  sendbuf, int *  sendcnts, int *  sdispls, T *  recvbuf, int *  recvcnts, int *  rdispls, MPI_Comm  comm )

Author: Rahul S. Sampath Definition at line 226 of file parUtils.txx. References PROF_A2AV_WAIT_BEGIN, PROF_A2AV_WAIT_END, PROF_PAR_ALL2ALLV_SPARSE_BEGIN, and PROF_PAR_ALL2ALLV_SPARSE_END.

template<typename T> int par::Mpi_Bcast (  T *  buffer, int  count, int  root, MPI_Comm  comm )  [inline]

Author: Rahul S. Sampath Definition at line 123 of file parUtils.txx. References PROF_PAR_BCAST_BEGIN, and PROF_PAR_BCAST_END.

template<typename T> int par::Mpi_Gather (  T *  sendBuffer, T *  recvBuffer, int  count, int  root, MPI_Comm  comm )  [inline]

Author: Rahul S. Sampath Definition at line 110 of file parUtils.txx. References PROF_PAR_GATHER_BEGIN, and PROF_PAR_GATHER_END.

template<typename T> int par::Mpi_Irecv (  T *  buf, int  count, int  source, int  tag, MPI_Comm  comm, MPI_Request *  request )  [inline]

Definition at line 50 of file parUtils.txx.

template<typename T> int par::Mpi_Issend (  T *  buf, int  count, int  dest, int  tag, MPI_Comm  comm, MPI_Request *  request )  [inline]

Definition at line 28 of file parUtils.txx.

template<typename T> int par::Mpi_Recv (  T *  buf, int  count, int  source, int  tag, MPI_Comm  comm, MPI_Status *  status )  [inline]

Definition at line 39 of file parUtils.txx.

template<typename T> int par::Mpi_Reduce (  T *  sendbuf, T *  recvbuf, int  count, MPI_Op  op, int  root, MPI_Comm  comm )  [inline]

Author: Rahul S. Sampath Definition at line 135 of file parUtils.txx. References PROF_PAR_REDUCE_BEGIN, and PROF_PAR_REDUCE_END.

template<typename T> int par::Mpi_Scan (  T *  sendbuf, T *  recvbuf, int  count, MPI_Op  op, MPI_Comm  comm )  [inline]

Author: Rahul S. Sampath Definition at line 73 of file parUtils.txx. References PROF_PAR_SCAN_BEGIN, and PROF_PAR_SCAN_END.

template<typename T, typename S> int par::Mpi_Sendrecv (  T *  sendBuf, int  sendCount, int  dest, int  sendTag, S *  recvBuf, int  recvCount, int  source, int  recvTag, MPI_Comm  comm, MPI_Status *  status )  [inline]

Author: Rahul S. Sampath Definition at line 61 of file parUtils.txx. References PROF_PAR_SENDRECV_BEGIN, and PROF_PAR_SENDRECV_END.

template<typename T> void par::Par_bitonic_merge_incr (  std::vector< T > &  local_list, int  proc_set_size, MPI_Comm  comm )

Author: Hari Sundar Definition at line 1571 of file parUtils.txx. References binOp::getPrevHighestPowerOfTwo(). Referenced by bitonicSort().

template<typename T> void par::Par_bitonic_sort_decr (  std::vector< T > &  local_list, int  proc_set_size, MPI_Comm  comm )

Author: Hari Sundar Definition at line 1539 of file parUtils.txx.

template<typename T> void par::Par_bitonic_sort_incr (  std::vector< T > &  local_list, int  proc_set_size, MPI_Comm  comm )

Author: Hari Sundar Definition at line 1507 of file parUtils.txx.

template<typename T> int par::partitionW (  std::vector< T > &  vec, unsigned int(*)(const T *)  getWeight, MPI_Comm  comm )

A parallel weighted partitioning function. In our implementation, we do not pose any restriction on the input or the number of processors. This function can be used with an odd number of processors as well. Some processors can pass an empty vector as input. The relative ordering of the elements is preserved. Author: Hari Sundar Rahul Sampath Parameters: vec  the input vector getWeight  function pointer to compute the weight of each element. If you pass NULL, then every element will get a weight equal to 1. comm  the communicator Definition at line 930 of file parUtils.txx. References DendroIntL, PROF_PARTW_BEGIN, and PROF_PARTW_END.

template<typename T> int par::removeDuplicates (  std::vector< T > &  nodes, bool  isSorted, MPI_Comm  comm )

Removes duplicates in parallel. If the input is not sorted, sample sort will be called within the function to sort the vector and then duplicates will be removed. Parameters: nodes  the input vector. isSorted  pass 'true' if the vector is globally sorted. comm  The communicator Author: Rahul Sampath Definition at line 1174 of file parUtils.txx. References PROF_REMDUP_BEGIN, PROF_REMDUP_END, and splitComm2way().

template<typename T> int par::sampleSort (  std::vector< T > &  in, std::vector< T > &  out, MPI_Comm  comm )

A parallel sample sort implementation. In our implementation, we do not pose any restriction on the input or the number of processors. This function can be used with an odd number of processors as well. Some processors can pass an empty vector as input. If the total number of elements in the vector (globally) is fewer than 10*p^2, where p is the number of processors, then we will use bitonic sort instead of sample sort to sort the vector. We use a paralle bitonic sort to sort the samples in the sample sort algorithm. Hence, the complexity of the algorithm is O(n/p log n/p) + O(p log p). Here, n is the global length of the vector and p is the number of processors. Author: Hari Sundar Rahul Sampath Santi Swaroop Adavani Shravan Veerapaneni Parameters: in  the input vector out  the output vector comm  the communicator Definition at line 1282 of file parUtils.txx. References DendroIntL, PROF_SORT_BEGIN, PROF_SORT_END, sort(), and splitCommUsingSplittingRank().

template<typename T> int par::scatterValues (  std::vector< T > &  in, std::vector< T > &  out, DendroIntL  outSz, MPI_Comm  comm )

Re-distributes a STL vector, preserving the relative ordering of the elements. Author: Rahul S. Sampath Parameters: in  The input vector out  The output vector. Memory for this vector will be allocated within the function. outSz  The local size of the output vector on the calling processor. comm  The communicator Returns: error flag Definition at line 398 of file parUtils.txx. References DendroIntL, PROF_PAR_SCATTER_BEGIN, and PROF_PAR_SCATTER_END. Referenced by ot::prolongMatVecType2(), and ot::restrictMatVecType2().

int par::splitComm2way (  const bool *  isEmptyList, MPI_Comm *  new_comm, MPI_Comm  orig_comm )

Splits a communication group into two depending on the values in isEmptyList. Both the groups are sorted in the ascending order of their ranks in the old comm. All processors must call this function with the same 'isEmptyList' array. Author: Rahul Sampath Parameters: isEmptyList  flags (of length equal to the number of processors) to determine whether each processor is active or not. orig_comm  The comm group that needs to be split. new_comm  The new comm group. Definition at line 216 of file parUtils.C.

int par::splitComm2way (  bool  iAmEmpty, MPI_Comm *  new_comm, MPI_Comm  orig_comm )

Splits a communication group into two, one containing processors that passed a value of 'false' for the parameter 'iAmEmpty' and the another containing processors that passed a value of 'true' for the parameter. Both the groups are sorted in the ascending order of their ranks in the old comm. Author: Rahul Sampath Parameters: iAmEmpty  Some flag to determine which group the calling processor will be combined into. orig_comm  The comm group that needs to be split. new_comm  The new comm group. Definition at line 114 of file parUtils.C. References PROF_SPLIT_COMM_2WAY_BEGIN, and PROF_SPLIT_COMM_2WAY_END. Referenced by ot::completeOctree(), ComputeFBM2_RHS_Part1(), ComputeFBM_RHS_Part2(), ComputeRHS9(), createGDmatType2_Type2(), createLmatType1_Type2(), createLmatType2_Type2(), createMmatType1_Type2(), createMmatType2_Type2(), createRHSType2_Type2(), ot::createShapeFnCoeffs_Type2(), createShFnMat_Type2(), ot::DA::DA_FactoryPart0(), ot::DA::DA_FactoryPart1(), ot::DAMG_InitPrivateType2(), ot::DAMGCreateAndSetDA(), par::test::isSorted(), par::test::isUniqueAndSorted(), ot::lineariseList(), main(), removeDuplicates(), and ot::simpleCoarsen().

unsigned int par::splitCommBinary (  MPI_Comm  orig_comm, MPI_Comm *  new_comm )

Splits a communication group into two, the first having a power of 2 number of processors and the other having the remainder. The first group is sorted in the ascending order of their ranks in the old comm and the second group is sorted in the descending order of their ranks in the old comm. Author: Hari Sundar Parameters: orig_comm  The comm group that needs to be split. new_comm  The new comm group. Definition at line 21 of file parUtils.C. References binOp::getPrevHighestPowerOfTwo(). Referenced by bitonicSort().

unsigned int par::splitCommBinaryNoFlip (  MPI_Comm  orig_comm, MPI_Comm *  new_comm )

Splits a communication group into two, the first having a power of 2 number of processors and the other having the remainder. Both the groups are sorted in the ascending order of their ranks in the old comm. Author: Hari Sundar Parameters: orig_comm  The comm group that needs to be split. new_comm  The new comm group. Definition at line 67 of file parUtils.C. References binOp::getPrevHighestPowerOfTwo(). Referenced by bitonicSort().

int par::splitCommUsingSplittingRank (  int  splittingRank, MPI_Comm *  new_comm, MPI_Comm  orig_comm )

Definition at line 172 of file parUtils.C. References PROF_SPLIT_COMM_BEGIN, and PROF_SPLIT_COMM_END. Referenced by ot::balanceOctree(), ot::coarsenOctree(), ot::DA::DA_FactoryPart3(), ot::DAMGCreateAndSetDA(), ot::points2Octree(), ot::regularGrid2Octree(), and sampleSort().

---