parallel_nodes.h

Go to the documentation of this file.

/*
 * Copyright (c) 2011-2015:  G-CSC, Goethe University Frankfurt
 * Author: Martin Rupp
 * 
 * This file is part of UG4.
 * 
 * UG4 is free software: you can redistribute it and/or modify it under the
 * terms of the GNU Lesser General Public License version 3 (as published by the
 * Free Software Foundation) with the following additional attribution
 * requirements (according to LGPL/GPL v3 §7):
 * 
 * (1) The following notice must be displayed in the Appropriate Legal Notices
 * of covered and combined works: "Based on UG4 (www.ug4.org/license)".
 * 
 * (2) The following notice must be displayed at a prominent place in the
 * terminal output of covered works: "Based on UG4 (www.ug4.org/license)".
 * 
 * (3) The following bibliography is recommended for citation and must be
 * preserved in all covered files:
 * "Reiter, S., Vogel, A., Heppner, I., Rupp, M., and Wittum, G. A massively
 *   parallel geometric multigrid solver on hierarchically distributed grids.
 *   Computing and visualization in science 16, 4 (2013), 151-164"
 * "Vogel, A., Reiter, S., Rupp, M., Nägel, A., and Wittum, G. UG4 -- a novel
 *   flexible software system for simulating pde based models on high performance
 *   computers. Computing and visualization in science 16, 4 (2013), 165-179"
 * 
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 * GNU Lesser General Public License for more details.
 */
 
#ifndef UG_PARALLEL_NODES_H_
#define UG_PARALLEL_NODES_H_
 
#include "pcl/pcl.h"
#include <vector>
#include <set>
#include <map>
 
#include "lib_algebra/parallelization/parallelization_util.h"
#include "lib_algebra/parallelization/algebra_layouts.h"
 
#include "common/util/sort_util.h"
#include "common/util/binary_buffer.h"
#include "common/serialization.h"
 
 
#define NAE_APPEND
//  serializes data from a vector<bool> into a binary stream
/*template <class TOStream>
void Serialize(TOStream& buf, const std::vector<bool>& vec)
{
  size_t size = vec.size();
  Serialize(buf, size);
  size_t sm8 = size%8;
  size_t s8 = size - sm8;
  for(size_t i=0; i<s8; i+=8)
  {
    char a=0;
    for(size_t j=0; j<8; j++)
      if(vec[i+j])
        a &= (1 << j);
    Serialize(buf, a);
  }
  if(sm8)
  {
    char a=0;
    for(size_t j=0; j<sm8; j++)
      if(vec[s8+j])
        a &= 1 << j;
    Serialize(buf, a);
  }
}
 
template <class TIStream>
void Deserialize(TIStream& buf, std::vector<bool> &vec)
{
  size_t size = 0;
  Deserialize(buf, size);
  vec.resize(size);
  size_t sm8 = size%8;
  size_t s8 = size - sm8;
  char a;
  for(size_t i=0; i<s8; i+=8)
  {
    Deserialize(buf, a);
    char a=0;
    for(size_t j=0; j<8; j++)
      vec[i+j] = (a & (1 << j));
  }
  if(sm8)
  {
    Deserialize(buf, a);
    for(size_t j=0; j<sm8; j++)
      vec[s8+j] = (a & (1 << j));
  }
}*/
 
namespace ug
{
 
class ParallelNodes
{
private:
  typedef std::map<AlgebraID,size_t>::iterator iterator;
  typedef std::map<AlgebraID,size_t>::const_iterator const_iterator;
 
public:
  struct OverlapType
  {
#define OT_SLAVE_FLAG 4
#define OT_MASTER_FLAG 2
#define OT_INNER_FLAG 1
 
    enum eOverlapType
    {
      OT_MASTER, OT_SLAVE, OT_OUTER
    };
    enum eOverlapType {…};
    OverlapType()
    {
      // create as inner node
      set_inner();
    }
    OverlapType() {…}
 
    OverlapType(int distanceToMasterOrInner)
    {
      set_distance_to_master_or_inner(distanceToMasterOrInner);
    }
    OverlapType(int distanceToMasterOrInner) {…}
 
    bool is_master_or_inner() const
    {
      return is_master() || is_inner();
    }
    bool is_master_or_inner() const {…}
 
    bool is_master() const
    {
      return type == OT_MASTER_FLAG;
    }
    bool is_master() const {…}
    bool is_slave() const
    {
      return type == (OT_SLAVE_FLAG & (1 << 3));
    }
    bool is_slave() const {…}
 
    bool is_inner() const
    {
      return type & OT_INNER_FLAG;
    }
    bool is_inner() const {…}
 
    void set_inner()
    {
      type = OT_INNER_FLAG;
    }
    void set_inner() {…}
 
    void set_master()
    {
      type = OT_MASTER_FLAG;
    }
    void set_master() {…}
 
    void set_slave()
    {
      type = OT_SLAVE_FLAG & (1 << 3);
    }
    void set_slave() {…}
 
    void set_distance_to_master_or_inner(size_t i)
    {
      type = (i << 3);
    }
    void set_distance_to_master_or_inner(size_t i) {…}
 
    size_t distance_to_master_or_inner() const
    {
      return type >> 3;
    }
    size_t distance_to_master_or_inner() const {…}
 
    friend std::ostream &operator << (std::ostream &out, const OverlapType &o)
    {
      if(o.is_master())
        out << "master ";
      if(o.is_slave())
        out << "slave ";
      if(o.is_inner())
        out << "inner ";
      out << "dTMI=" << o.distance_to_master_or_inner();
      return out;
    }
    friend std::ostream &operator << (std::ostream &out, const OverlapType &o) {…};
 
    int type;
  };
  struct OverlapType {…};
private:
  struct NewSlaveNotification
  {
    NewSlaveNotification() : id() {}
    NewSlaveNotification(const AlgebraID &_id, int _newSlaveOnPID)
    {
      id = _id;
      newSlaveOnPID = _newSlaveOnPID;
    }
    NewSlaveNotification(const AlgebraID &_id, int _newSlaveOnPID) {…}
    friend std::ostream &operator << (std::ostream &out, const NewSlaveNotification &n)
    {
      out << "notification that node " << n.id << " is slave on " << n.newSlaveOnPID << " ";
      return out;
    }
    friend std::ostream &operator << (std::ostream &out, const NewSlaveNotification &n) {…};
    AlgebraID id;
    int newSlaveOnPID;
  };
  struct NewSlaveNotification {…};
 
private:
  ConstSmartPtr<AlgebraLayouts> m_layout;
 
  std::map<AlgebraID, size_t> m_globalToLocal;
  std::vector<AlgebraID> m_localToGlobal;
 
  IndexLayout totalMasterLayout;
  IndexLayout totalSlaveLayout;
 
  size_t m_originalSize;
  std::vector<OverlapType> m_OLtype;
 
  typedef std::map<int, BinaryBuffer> BufferMap;
 
  std::set<int> masterPIDs;
  std::set<int> slavePIDs;
 
  BufferMap notificationBufferMap;
 
 
  std::map<int, std::set<size_t> > notified; // notified[pid] is the set of indices which are slave on pid.
  std::map<int, std::vector<NewSlaveNotification> > newSlaveNotifications; // newSlaveNotifications is a vector of pids
  std::map<int, std::set<size_t> > newMasters;
  std::map<int, std::set<size_t> > newSlaves;
 
public:
  ParallelNodes();
  ParallelNodes(const ParallelNodes &);
  ParallelNodes(ConstSmartPtr<AlgebraLayouts> layout, size_t s);
 
  size_t get_original_size()
  {
    return m_originalSize;
  }
  size_t get_original_size() {…}
 
  size_t local_size() const
  {
    return m_localToGlobal.size();
  }
  size_t local_size() const {…}
 
  const AlgebraID &local_to_global(size_t i) const
  {
    UG_COND_THROW(i >= local_size(), i << " >= " << local_size());
    return m_localToGlobal[i];
  }
  const AlgebraID &local_to_global(size_t i) const {…}
 
  const AlgebraID &operator [] (size_t i) const
  {
    return local_to_global(i);
  }
  const AlgebraID &operator [] (size_t i) const {…}
 
  size_t get_local_index_if_available(const AlgebraID &globalIndex, bool &bHasIndex) const;
 
  size_t get_local_index_or_create_new(const AlgebraID &globalIndex, int distanceToMasterOrInner, bool &bCreated);
 
 
  size_t get_local_index_or_create_new(const AlgebraID &globalIndex, int distanceToMasterOrInner);
 
  size_t global_to_local(const AlgebraID &globalIndex) const;
 
  size_t operator [] (const AlgebraID &globalIndex) const
  {
    return global_to_local(globalIndex);
  }
  size_t operator [] (const AlgebraID &globalIndex) const {…}
 
 
  void sort_by_global_id(std::vector<size_t> &v)
  {
    sort(v.begin(), v.end(), CompareIndicesBy(m_localToGlobal) );
  }
  void sort_by_global_id(std::vector<size_t> &v) {…}
 
  void insert_into_interface_sorted(std::vector<size_t> &v, IndexLayout::Interface &interface);
 
  void insert_into_layout_sorted(std::map<int, std::set<size_t> > &m, IndexLayout &layout);
 
  void insert_into_layout_sorted(std::map<int, std::vector<size_t> > &m, IndexLayout &layout);
 
  void sort_interface(IndexLayout::Interface &interface);
 
  void sort_layout(IndexLayout &layout);
 
 
  IndexLayout &get_total_master_layout()
  {
    return totalMasterLayout;
  }
  IndexLayout &get_total_master_layout() {…}
  IndexLayout &get_total_slave_layout()
  {
    return totalSlaveLayout;
  }
  IndexLayout &get_total_slave_layout() {…}
 
  const IndexLayout &master_layout() const
  {
    return m_layout->master();
  }
  const IndexLayout &master_layout() const {…}
 
  const IndexLayout &slave_layout() const
  {
    return m_layout->slave();
  }
  const IndexLayout &slave_layout() const {…}
 
  pcl::InterfaceCommunicator<IndexLayout> &comm() const
  {
    return m_layout->comm();
  }
  pcl::InterfaceCommunicator<IndexLayout> &comm() const {…}
 
  const pcl::ProcessCommunicator &proc_comm() const
  {
    return m_layout->proc_comm();
  }
  const pcl::ProcessCommunicator &proc_comm() const {…}
 
  bool is_master_or_inner(size_t i) const
  {
    return m_OLtype[i].is_master_or_inner();
  }
  bool is_master_or_inner(size_t i) const {…}
 
  bool is_inner(size_t i) const
  {
    return m_OLtype[i].is_inner();
  }
  bool is_inner(size_t i) const {…}
  bool is_slave(size_t i) const
  {
    return m_OLtype[i].is_slave();
  }
  bool is_slave(size_t i) const {…}
  bool is_master(size_t i) const
  {
    return m_OLtype[i].is_master();
  }
  bool is_master(size_t i) const {…}
  size_t distance_to_master_or_inner(size_t i) const
  {
    return m_OLtype[i].distance_to_master_or_inner();
  }
  size_t distance_to_master_or_inner(size_t i) const {…}
  void print() const
  {
    for(size_t i=0; i<local_size(); i++)
      UG_LOG(i << ": " << m_OLtype[i] << " globalID = " << local_to_global(i) << "\n");
 
  }
  void print() const {…}
 
  const OverlapType &overlap_type(size_t i)
  {
    return m_OLtype[i];
  }
  const OverlapType &overlap_type(size_t i) {…}
 
  size_t create_slave_node(const AlgebraID &globalID, int distanceToMasterOrInner);
 
  void create_node(const AlgebraID &globalID, size_t localIndex, int pid);
 
  void create_node(const AlgebraID &globalID, int pid)
  {
    create_node(globalID, global_to_local(globalID), pid);
  }
  void create_node(const AlgebraID &globalID, int pid) {…}
 
  void create_node(size_t localIndex, int pid)
  {
    create_node(local_to_global(localIndex), localIndex, pid);
  }
  void create_node(size_t localIndex, int pid) {…}
 
 
  void issue(pcl::InterfaceCommunicator<IndexLayout> &communicator);
 
  void process();
 
  void add_new_layouts_to(IndexLayout &newMasterLayout, IndexLayout &newSlaveLayout);
#ifdef NAE_APPEND
  void append_nodes_without_comm(size_t n);
#endif
private:
  void create_mark_map(const IndexLayout &masterLayout);
 
  /*void check_new_nodes(pcl::InterfaceCommunicator<IndexLayout> &communicator,
        IndexLayout &sendingLayout, IndexLayout &receivingLayout)
  {
    std::set<int> sendingPIDs;
    for(IndexLayout::iterator it = sendingLayout.begin(); it != sendingLayout.end(); ++it)
      sendingPIDs.insert(sendingLayout.proc_id(it));
 
    std::set<int> receivingPIDs;
    for(IndexLayout::iterator it = receivingLayout.begin(); it != receivingLayout.end(); ++it)
      receivingPIDs.insert(receivingLayout.proc_id(it));
    check_new_nodes(commnunicator, sendingPIDs, receivingPIDs);
  }
 
  void check_new_nodes(std::vector<int> sendingPIDs, std::vector<int> receivingPIDs)
  {
    for(size_t i=0; i<sendingPIDs.size(); i++)
    {
      BinaryBuffer buf;
      int pid = sendingPIDs[i];
      Serialize(buf, newNodesOn[pid]);
      communicator.send_raw(pid, buf.buffer(), buf.write_pos(), false);
    }
 
    for(size_t i=0; i<receivingPIDs.size(); i++)
      communicator.receive_raw(pid, newNodesBufferMap[receivingPIDs[i]]);
 
    communicator.communicate();
  }*/
 
 
};
class ParallelNodes {…};
 
 
}
#endif