inbufastric.hpp

// ***********************************************************************
//
//                              TRIC
//
// ***********************************************************************
//
//       Copyright (2019) Battelle Memorial Institute
//                      All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions
// are met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the copyright holder nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
// FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
// COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
// INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
// BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
// LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
// CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
// LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
// ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
// POSSIBILITY OF SUCH DAMAGE.
//
// ************************************************************************ 
#pragma once
#ifndef INBUFASTRIC_HPP
#define INBUFASTRIC_HPP

#include "graph.hpp"

#include <numeric>
#include <utility>
#include <cstring>
#include <iomanip>
#include <limits>
#include <cmath>
#include <vector>
#include <stdexcept>
#include <string> 

#ifndef TAG_DATA
#define TAG_DATA  (100)
#endif

class TriangulateAggrBufferedInrecv
{
  public:

    TriangulateAggrBufferedInrecv(Graph* g, const GraphElem bufsize): 
      g_(g), sbuf_ctr_(nullptr), sbuf_(nullptr), rbuf_(nullptr), 
      pdegree_(0), sreq_(nullptr), erange_(nullptr), vcount_(nullptr), ntriangles_(0), 
      nghosts_(0), out_nghosts_(0), in_nghosts_(0), pindex_(0), prev_m_(nullptr), 
      prev_k_(nullptr), stat_(nullptr), targets_(0), sources_(0), bufsize_(0), rreq_(nullptr),
      recv_count_(nullptr), rindex_(0), rdegree_(0)
#if defined(RECV_MAP)
      , rmap_({})
#endif
  {
    comm_ = g_->get_comm();
    MPI_Comm_size(comm_, &size_);
    MPI_Comm_rank(comm_, &rank_);

    const GraphElem lnv = g_->get_lnv();
    const GraphElem nv = g_->get_nv();

    vcount_ = new GraphElem[lnv]();
    erange_ = new GraphElem[nv*2]();
    GraphElem *send_count  = new GraphElem[size_]();
    recv_count_  = new GraphElem[size_]();

    double t0 = MPI_Wtime();

    // store edge ranges
    GraphElem base = g_->get_base(rank_);
    for (GraphElem i = 0; i < lnv; i++)
    {
      GraphElem e0, e1;
      g_->edge_range(i, e0, e1);

      if ((e0 + 1) == e1)
        continue;

      Edge const& edge_s = g_->get_edge(e0);
      Edge const& edge_t = g_->get_edge(e1-1);

      erange_[(i + base)*2] = edge_s.tail_;
      erange_[(i + base)*2+1] = edge_t.tail_;
    }
    
    MPI_Barrier(comm_);
    
    MPI_Allreduce(MPI_IN_PLACE, erange_, nv*2, MPI_GRAPH_TYPE, 
        MPI_SUM, comm_);

    for (GraphElem i = 0; i < lnv; i++)
    {
      GraphElem e0, e1, tup[2];
      g_->edge_range(i, e0, e1);

      if ((e0 + 1) == e1)
        continue;

      for (GraphElem m = e0; m < e1-1; m++)
      {
        Edge const& edge_m = g_->get_edge(m);
        const int owner = g_->get_owner(edge_m.tail_);
        tup[0] = edge_m.tail_;
        if (owner == rank_)
        {
          for (GraphElem n = m + 1; n < e1; n++)
          {
            Edge const& edge_n = g_->get_edge(n);

            tup[1] = edge_n.tail_;
            if (fast_check_edgelist(tup))
              ntriangles_ += 1;
          }
        }
        else
        {
          for (GraphElem n = m + 1; n < e1; n++)
          {
            Edge const& edge_n = g_->get_edge(n);
             
            if (!edge_within_max(edge_m.tail_, edge_n.tail_))
              break;
            
            if (!edge_above_min(edge_m.tail_, edge_n.tail_))
              continue;

            send_count[owner] += 1;
            vcount_[i] += 1;
          }
        }
      }
    }

    MPI_Barrier(comm_);

    double t1 = MPI_Wtime();
    double p_tot = t1 - t0, t_tot = 0.0;

    MPI_Reduce(&p_tot, &t_tot, 1, MPI_DOUBLE, MPI_SUM, 0, comm_);

    if (rank_ == 0) 
    {   
      std::cout << "Average time for local counting during instantiation (secs.): " 
        << ((double)(t_tot / (double)size_)) << std::endl;
    }

    t0 = MPI_Wtime();

    // outgoing/incoming data and buffer size
    MPI_Alltoall(send_count, 1, MPI_GRAPH_TYPE, recv_count_, 1, MPI_GRAPH_TYPE, comm_);
    
    for (GraphElem p = 0; p < size_; p++)
    {
      out_nghosts_ += send_count[p];
      in_nghosts_ += recv_count_[p];

      if (send_count[p] > 0)
        targets_.push_back(p);

      if (recv_count_[p] > 0)
        sources_.push_back(p);
    }
      
    pdegree_ = targets_.size();
    rdegree_ = sources_.size();

    for (int i = 0; i < pdegree_; i++)
      pindex_.insert({targets_[i], i});

    for (int i = 0; i < rdegree_; i++)
      rindex_.insert({sources_[i], i});

    nghosts_ = out_nghosts_ + in_nghosts_;

    bufsize_ = ((nghosts_*2) < bufsize) ? (nghosts_*2) : bufsize;

    MPI_Allreduce(MPI_IN_PLACE, &bufsize_, 1, MPI_GRAPH_TYPE, MPI_MAX, comm_);

    if (rank_ == 0)
      std::cout << "Adjusted Per-PE buffer count: " << bufsize_ << std::endl;
 
    sbuf_     = new GraphElem[pdegree_*bufsize_];
    sbuf_ctr_ = new GraphElem[pdegree_]();
    prev_k_   = new GraphElem[pdegree_];
    prev_m_   = new GraphElem[pdegree_];
    sreq_     = new MPI_Request[pdegree_];
    stat_     = new char[pdegree_];
    
    rbuf_     = new GraphElem[rdegree_*bufsize_];
    rreq_     = new MPI_Request[rdegree_];

    std::fill(sreq_, sreq_ + pdegree_, MPI_REQUEST_NULL);
    std::fill(prev_k_, prev_k_ + pdegree_, -1);
    std::fill(prev_m_, prev_m_ + pdegree_, -1);
    std::fill(stat_, stat_ + pdegree_, '0');
    
    std::fill(rreq_, rreq_ + rdegree_, MPI_REQUEST_NULL);

    MPI_Barrier(comm_);

#if defined(DEBUG_PRINTF)
    if (rank_ == 0)
    {
      std::cout << "Edge range per vertex (#ID: <range>): " << std::endl;
      for (int i = 0, j = 0; i < nv*2; i+=2, j++)
        std::cout << j << ": " << erange_[i] << ", " << erange_[i+1] << std::endl;
    }
#endif
    
    delete []send_count;
  }

    ~TriangulateAggrBufferedInrecv() {}

    void clear()
    {
      delete []sbuf_;
      delete []rbuf_;
      delete []recv_count_;
      delete []sbuf_ctr_;
      delete []sreq_;
      delete []rreq_;
      delete []prev_k_;
      delete []prev_m_;
      delete []stat_;
      delete []vcount_;
      delete []erange_;

#if defined(RECV_MAP)
      rmap_.clear();
#endif

      pindex_.clear();
      targets_.clear();
      rindex_.clear();
      sources_.clear();
    }

    inline void nbsend(int owner)
    {
      if (stat_[pindex_[owner]] == '0' && sbuf_ctr_[pindex_[owner]] > 0)
      {
        MPI_Isend(&sbuf_[pindex_[owner]*bufsize_], sbuf_ctr_[pindex_[owner]], 
            MPI_GRAPH_TYPE, owner, TAG_DATA, comm_, &sreq_[pindex_[owner]]);
        stat_[pindex_[owner]] = '1'; 
      }
    }

    inline void nbsend()
    {
#if defined(USE_OPENMP) && !defined(RECV_MAP)
#pragma omp parallel for default(shared)
#endif
      for (int p = 0; p < pdegree_; p++)
        nbsend(targets_[p]);
    }

    inline void nbrecv(int source)
    {
        if (recv_count_[source] > 0)
        {
          MPI_Irecv(&rbuf_[rindex_[source]*bufsize_], bufsize_, 
              MPI_GRAPH_TYPE, source, TAG_DATA, comm_, &rreq_[rindex_[source]]);
        }
    }

    inline void nbrecv()
    {
#if defined(USE_OPENMP) && !defined(RECV_MAP) 
#pragma omp parallel for default(shared)
#endif
      for (int p = 0; p < rdegree_; p++)
        nbrecv(sources_[p]);
    }
    
    inline void lookup_edges()
    {
      const GraphElem lnv = g_->get_lnv();

      for (GraphElem i = 0; i < lnv; i++)
      {
        if (vcount_[i] == 0) // all edges processed, move on
          continue;

        GraphElem e0, e1;
        g_->edge_range(i, e0, e1);

        if ((e0 + 1) == e1)
          continue;

        for (GraphElem m = e0; m < e1-1; m++)
        {
          EdgeStat& edge = g_->get_edge_stat(m);
          const int owner = g_->get_owner(edge.edge_->tail_);
          const GraphElem pidx = pindex_[owner];
          const GraphElem disp = pidx*bufsize_;

          if (owner != rank_ && edge.active_)
          {   
            if (stat_[pidx] == '1') 
              continue;

            if (m >= prev_m_[pidx])
            {
              if (sbuf_ctr_[pidx] == (bufsize_-1))
              {
                prev_m_[pidx] = m;
                prev_k_[pidx] = -1;

                nbsend(owner);

                continue;
              }

              sbuf_[disp+sbuf_ctr_[pidx]] = edge.edge_->tail_;
              sbuf_ctr_[pidx] += 1;

              for (GraphElem n = ((prev_k_[pidx] == -1) ? (m + 1) : prev_k_[pidx]); n < e1; n++)
              {  
                Edge const& edge_n = g_->get_edge(n);                                
                                  
                if (!edge_within_max(edge.edge_->tail_, edge_n.tail_))
                  break;
                
                if (sbuf_ctr_[pidx] == (bufsize_-1))
                {
                  prev_m_[pidx] = m;
                  prev_k_[pidx] = n;

                  sbuf_[disp+sbuf_ctr_[pidx]] = -1; // demarcate vertex boundary
                  sbuf_ctr_[pidx] += 1;

                  nbsend(owner);

                  break;
                }
                
                if (!edge_above_min(edge.edge_->tail_, edge_n.tail_))
                  continue;
             
                sbuf_[disp+sbuf_ctr_[pidx]] = edge_n.tail_;
                sbuf_ctr_[pidx] += 1;
                
                out_nghosts_ -= 1;
                vcount_[i] -= 1;
              }
              
              if (stat_[pidx] == '0') 
              {               
                prev_m_[pidx] = m;
                prev_k_[pidx] = -1;
                
                edge.active_ = false;

                if (sbuf_ctr_[pidx] == (bufsize_-1))
                {
                  sbuf_[disp+sbuf_ctr_[pidx]] = -1; 
                  sbuf_ctr_[pidx] += 1;

                  nbsend(owner);
                }
                else
                {
                  sbuf_[disp+sbuf_ctr_[pidx]] = -1; 
                  sbuf_ctr_[pidx] += 1;
                }
              }
            }
          }
        }
      }
    }

    inline bool check_edgelist(GraphElem (&tup)[2])
    {
      GraphElem e0, e1;
      const GraphElem lv = g_->global_to_local(tup[0]);
      g_->edge_range(lv, e0, e1);
      for (GraphElem e = e0; e < e1; e++)
      {
        Edge const& edge = g_->get_edge(e);
        if (edge.tail_ > tup[1])
          break;
        if (tup[1] == edge.tail_)
          return true;
      }
      return false;
    }
    
    inline bool fast_check_edgelist(GraphElem (&tup)[2])
    {
      GraphElem e0, e1;
      const GraphElem lv = g_->global_to_local(tup[0]);
      
      g_->edge_range(lv, e0, e1);
      Edge const& first_edge = g_->get_edge(e0);
      Edge const& last_edge = g_->get_edge(e1 - 1);

      if (tup[1] == last_edge.tail_ || tup[1] == first_edge.tail_)
        return true;
      
      if (tup[1] > first_edge.tail_ && tup[1] < last_edge.tail_)
      {
        Edge const& mid_edge = g_->get_edge(std::midpoint(e0, e1));

        if (tup[1] == mid_edge.tail_)
          return true;
        else if (tup[1] > mid_edge.tail_)
        {
          for (GraphElem e = std::midpoint(e0, e1) + 1; e < e1; e++)
          {
            Edge const& edge = g_->get_edge(e);
            if (edge.tail_ > tup[1])
              break;
            if (tup[1] == edge.tail_)
              return true;
          }
        }
        else
        {
          for (GraphElem e = e0; e < std::midpoint(e0, e1); e++)
          {
            Edge const& edge = g_->get_edge(e);
            if (edge.tail_ > tup[1])
              break;
            if (tup[1] == edge.tail_)
              return true;
          }
        }
      }
        
      return false;
    }

#if defined(USE_OPENMP_NESTED)
    inline bool check_edgelist_omp(GraphElem (&tup)[2])
    {
      GraphElem e0, e1;
      bool found = false;
      
      const GraphElem lv = g_->global_to_local(tup[0]);
      g_->edge_range(lv, e0, e1);

#pragma omp parallel for schedule(dynamic) firstprivate(tup) shared(g_, e0, e1) reduction(||:found) if (e1 - e0 >= 50)
      for (GraphElem e = e0; e < e1; e++)
      {
        Edge const& edge = g_->get_edge(e);
        
        if (tup[1] == edge.tail_)
          found = found || true;
      }

      return found;
    }
#endif

    inline bool edge_between_range(GraphElem const& x, GraphElem const& y) const
    {
      if ((y >= erange_[x*2]) && (y <= erange_[x*2+1]))
        return true;
      return false;
    }
         
    inline bool edge_above_min(GraphElem x, GraphElem y) const
    {
      if (y >= erange_[x*2] || x >= erange_[y*2])
        return true;
      return false;
    }

    inline bool edge_within_max(GraphElem x, GraphElem y) const
    {
      if (y <= erange_[x*2+1] || x <= erange_[y*2+1])
        return true;
      return false;
    }
   
#if defined(RECV_MAP)
    bool find_pair(GraphElem (&pair)[2])
    {
      if (rmap_.count(pair[0]))
      {
        typedef std::multimap<const GraphElem, GraphElem>::const_iterator mmap_t;
        std::pair<mmap_t, mmap_t> range = rmap_.equal_range(pair[0]);
        
        for (mmap_t it = range.first; it != range.second; ++it)
          if (it->second == pair[1])
            return true;
      }
      return false;
    }

    void check_rmap(GraphElem (&pair)[2])
    {
      if (find_pair(pair)) 
        ntriangles_ += 1;
      else
      {
        if (check_edgelist(pair))
        {
          ntriangles_ += 1;
          rmap_.insert(std::pair<const GraphElem, GraphElem>{pair[0], pair[1]});
        }
      }
    }
#endif

    inline void process_recvs()
    {
      if (in_nghosts_ == 0)
        return;

#if defined(USE_OPENMP) && !defined(RECV_MAP)
#pragma omp parallel for default(shared) reduction(+:ntriangles_) reduction(-:in_nghosts_)
#endif
      for (int p = 0; p < rdegree_; p++)
      {
        int over = -1;
        MPI_Status status;

        MPI_Test(&rreq_[p], &over, &status);

        if (over)
        { 
          GraphElem tup[2] = {-1,-1}, prev = 0;
          int count = 0;

          const int source = status.MPI_SOURCE;

          MPI_Get_count(&status, MPI_GRAPH_TYPE, &count);

          if (source != MPI_ANY_SOURCE)
          {
            for (GraphElem k = 0; k < count;)
            {
              if (rbuf_[p*bufsize_+k] == -1)
              {
                k += 1;
                prev = k;
                continue;
              }

              tup[0] = rbuf_[p*bufsize_+k];
              GraphElem curr_count = 0;

              for (GraphElem m = k + 1; m < count; m++)
              {
                if (rbuf_[p*bufsize_+m] == -1)
                {
                  curr_count = m + 1;
                  break;
                }

                tup[1] = rbuf_[p*bufsize_+m];

#if defined(RECV_MAP)
                check_rmap(tup);
#else
#if defined(USE_OPENMP_NESTED)
                if (check_edgelist_omp(tup))
                  ntriangles_ += 1;
#else
                if (fast_check_edgelist(tup))
                  ntriangles_ += 1;
#endif
#endif
                in_nghosts_ -= 1;
                recv_count_[source] -= 1;
              }

              k += (curr_count - prev);
              prev = k;
            }

            if (recv_count_[source] > 0)
            {
              MPI_Irecv(&rbuf_[p*bufsize_], bufsize_, 
                  MPI_GRAPH_TYPE, source, TAG_DATA, comm_, &rreq_[p]);
            }
          }
        }
      }
    }

    inline GraphElem count()
    {
#if defined(USE_ALLREDUCE_FOR_EXIT)
      GraphElem count;
#else      
      bool done = false, nbar_active = false; 
      MPI_Request nbar_req = MPI_REQUEST_NULL;
#endif

      int* inds = new int[pdegree_]();
      int over = -1;
      
      nbrecv();

#if defined(USE_ALLREDUCE_FOR_EXIT)
      while(1)
#else
      while(!done)
#endif
      {
        process_recvs();

        if (out_nghosts_ == 0)
          nbsend();
        else
          lookup_edges();
        
        MPI_Testsome(pdegree_, sreq_, &over, inds, MPI_STATUSES_IGNORE);

        if (over > 0)
        {
          for (int i = 0; i < over; i++)
          {
            sbuf_ctr_[inds[i]] = 0;
            stat_[inds[i]] = '0';
          }
        }

#if defined(USE_ALLREDUCE_FOR_EXIT)
        count = in_nghosts_;
        MPI_Allreduce(MPI_IN_PLACE, &count, 1, MPI_GRAPH_TYPE, MPI_SUM, comm_);
        if (count == 0)
          break;
#else       
        if (nbar_active)
        {
          int test_nbar = -1;
          MPI_Test(&nbar_req, &test_nbar, MPI_STATUS_IGNORE);
          done = !test_nbar ? false : true;
        }
        else
        {
          if (in_nghosts_ == 0)
          {
            MPI_Ibarrier(comm_, &nbar_req);
            nbar_active = true;
          }
        }
#endif
#if defined(DEBUG_PRINTF)
        std::cout << "in/out: " << in_nghosts_ << ", " << out_nghosts_ << std::endl;
#endif
      }

      GraphElem ttc = 0, ltc = ntriangles_;
      MPI_Barrier(comm_);
      MPI_Reduce(&ltc, &ttc, 1, MPI_GRAPH_TYPE, MPI_SUM, 0, comm_);
      
      delete []inds;

      return (ttc/3);
    }

  private:
    Graph* g_;

    GraphElem ntriangles_, bufsize_, nghosts_, out_nghosts_, in_nghosts_, pdegree_, rdegree_;
    GraphElem *sbuf_, *rbuf_, *recv_count_, *prev_k_, *prev_m_, *sbuf_ctr_, *vcount_, *erange_;
    MPI_Request *sreq_, *rreq_;
    char *stat_;
    
    std::vector<int> targets_, sources_;
    int rank_, size_;
    std::unordered_map<int, int> pindex_, rindex_; 
#if defined(RECV_MAP)
    std::multimap<const GraphElem, GraphElem> rmap_;
#endif
    MPI_Comm comm_;
};
#endif