Multiplier.h

Go to the documentation of this file.

#ifndef _MULTIPLIER_H_
#define _MULTIPLIER_H_
 
#include "CombBLAS/CombBLAS.h"
#include "CCGrid.h"
#include "SUMMALayer.h"
 
namespace combblas {
 
template <typename IT, typename NT>
SpDCCols<IT, NT>* multiply(SpDCCols<IT, NT> & splitA, SpDCCols<IT, NT> & splitB, CCGrid & CMG, bool isBT, bool threaded)
{
    
    comm_bcast = 0, comm_reduce = 0, comp_summa = 0, comp_reduce = 0, comp_result =0, comp_reduce_layer=0;
    int myrank;
    MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
    std::vector< SpTuples<IT,NT>* > unreducedC;
    
    MPI_Barrier(MPI_COMM_WORLD);
    double time_beg = MPI_Wtime();
    
    SUMMALayer(splitA, splitB, unreducedC, CMG, isBT, threaded);
    
    MPI_Barrier(MPI_COMM_WORLD);
    double time_mid = MPI_Wtime();
    
    SpDCCols<IT,NT> * mergedC;
    mergedC = ReduceAll_threaded(unreducedC, CMG);
    MPI_Barrier(MPI_COMM_WORLD);
    double time_end = MPI_Wtime();
    double time_total = time_end-time_beg;
    
    /*
    int64_t local_nnz = mergedC->getnnz();
    int64_t global_nnz = 0;
    
    MPI_Reduce(&local_nnz, &global_nnz, 1, MPIType<int64_t>(), MPI_SUM, 0, MPI_COMM_WORLD);
    if(myrank == 0)
    {
        cout << "Global nonzeros in C is " << global_nnz << endl;
    }
     */
    
    int nthreads;
#pragma omp parallel
    {
        nthreads = omp_get_num_threads();
    }
    if(CMG.myrank == 0)
    {
        double time_other = time_total - (comm_bcast + comm_reduce + comp_summa + comp_reduce + comp_reduce_layer + comp_result);
        //printf(" ----------------------------------------------------------------------------------------------\n");
        //printf(" comm_bcast   comm_scatter comp_summa comp_merge  comp_scatter  comp_result     other      total\n");
        //printf(" ----------------------------------------------------------------------------------------------\n");
        
        //printf("%10lf %12lf %12lf %10lf %12lf %12lf %12lf %10lf\n\n", comm_bcast, comm_reduce, comp_summa, comp_reduce, comp_reduce_layer, comp_result, time_other, time_total);
        printf("%4d %4d %5d %6d %10lf %12lf %12lf %10lf %12lf %12lf %12lf %10lf\n", CMG.GridRows, CMG.GridCols, CMG.GridLayers, nthreads, comm_bcast, comm_reduce, comp_summa, comp_reduce, comp_reduce_layer, comp_result, time_other, time_total);
    }
    
    return mergedC;
}
 
}
 
#endif