void Histo::run()
{
  int me,nprocs;
  MPI_Comm_rank(MPI_COMM_WORLD,&me);
  MPI_Comm_size(MPI_COMM_WORLD,&nprocs);

  // MR = key : NULL

  MapReduce *mr = obj->input(1);
  uint64_t ntotal = mr->kv_stats(0);

  // unique keys and their count
  // before processing it, make a copy of input MR if it is permanent

  if (obj->permanent(mr)) mr = obj->copy_mr(mr);
  mr->collate(NULL);
  uint64_t nunique = mr->reduce(count,NULL);

  obj->output(1,mr);

  // histogram stats
  // before processing it, make a copy of output MR if it is permanent

  if (obj->permanent(mr)) mr = obj->copy_mr(mr);

  mr->map(mr,invert,NULL);
  mr->collate(NULL);
  mr->reduce(count,NULL);
  mr->gather(1);
  mr->sort_keys(-1);

  char msg[128];
  sprintf(msg,"Histo: %lu total keys, %lu unique",ntotal,nunique);
  if (me == 0) error->message(msg);
  mr->scan(print,NULL);

  obj->cleanup();
}
Ejemplo n.º 2
0
int MR_sort_keys(void *MRptr, int (*mycompare)(char *, int, char *, int))
{
  MapReduce *mr = (MapReduce *) MRptr;
  return mr->sort_keys(mycompare);
}
Ejemplo n.º 3
0
int main(int narg, char **args)
{
  int me, nprocs;
  int nwords, nunique;
  double tstart, tstop;
  Count count;
  pthread_t thread1;
  MPI_Init(&narg, &args);
  MPI_Comm_rank(MPI_COMM_WORLD, &me);
  MPI_Comm_size(MPI_COMM_WORLD, &nprocs);

  if (narg <= 1)
  {
    if (me == 0)
    {
      printf("Syntax: cwordfreq file1 file2 ...\n");
    }

    MPI_Abort(MPI_COMM_WORLD, 1);
  }

  if (me != 0)
  {
    pthread_create(&thread1, NULL, hb_function, &me);
  }
  else
  {
    printf("PID %d on ready for attach\n", getpid());
    pthread_create(&thread1, NULL, pingRecv, &me);
  }

  MapReduce *mr = new MapReduce(MPI_COMM_WORLD);
  MapReduce *mr2 = new MapReduce(MPI_COMM_WORLD);
  //MapReduce *mra = new MapReduce(MPI_COMM_WORLD);
  //MapReduce *mrb = new MapReduce(MPI_COMM_WORLD);
  MapReduce *mra = new MapReduce(MPI_COMM_WORLD);
  MapReduce *mrb = new MapReduce(MPI_COMM_WORLD);
//  mra->open();
  mr2->open();
  MPI_Barrier(MPI_COMM_WORLD);
  int test = 1;
  void *arg[5] = {args[1], args[2], mra->kv, mrb->kv, mr2->kv};
  //printf("%s %s\n", arg[0], arg[1]);
  printf("starting map\n");
  nwords = mr->map(nprocs, &fileread, arg);
  MPI_Barrier(MPI_COMM_WORLD);
  printf("map done\n");
  mr2->close();
//  mrb->close();
  printf("%d mr closed\n", me);
  mr->collate(NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  printf("collate done\n");
  nunique = mr->reduce(&sum2, NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  printf("reduce done\n");
  mr->gather(1);
  MPI_Barrier(MPI_COMM_WORLD);
  mr2->collate(NULL);
  mr2->reduce(&sum,NULL);
  mr2->gather(1);
  mr->add(mr2);
  mr->sort_keys(&ncompare);
  MPI_Barrier(MPI_COMM_WORLD);
  FILE * pFile;
  pFile = fopen("result.out", "w");
  mr->map(mr, &output, pFile);
  MPI_Barrier(MPI_COMM_WORLD);
  fclose(pFile);
//sum done
  //getHistogram(mra, "hist.a");
  //getHistogram(mr, "hist.a");
  //getHistogram(mrb, "hist.b");
  /* FILE * pFilea;
  printf("sum done\n");
   pFilea = fopen("test", "w");
   mra->gather(1);
   mra->map(mra, &histoutput, pFilea);
   fclose(pFilea);*/
  printf("%d sum done\n", me);
  mr->collate(NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  mr->reduce(&sum, NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  mr->gather(1);
  MPI_Barrier(MPI_COMM_WORLD);
  mr->sort_keys(&ncompare);
  MPI_Barrier(MPI_COMM_WORLD);
  FILE * pFile2;
  pFile2 = fopen("hist.c", "w");
  mr->map(mr, &histoutput, pFile2);
  MPI_Barrier(MPI_COMM_WORLD);
  fclose(pFile2);
  delete mr;
  printf("%d c done\n", me);
  mra->open();
  mrb->open();
  void *arg2[4] = {args[1], args[2], mra->kv, mrb->kv};
  printf("%p %p\n", mra->kv, arg2[2]);
  nwords = mra->map(nprocs, &fileread2, arg2);
  mra->close();
  mrb->close();
  mra->collate(NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  mra->reduce(&sum, NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  mra->gather(1);
  MPI_Barrier(MPI_COMM_WORLD);
  mra->sort_keys(&ncompare);
  MPI_Barrier(MPI_COMM_WORLD);
  mra->gather(1);
  MPI_Barrier(MPI_COMM_WORLD);
  FILE * pFile3;
  pFile3 = fopen("hist.a", "w");
  mra->map(mra, &histoutput, pFile3);
  fclose(pFile3);
  delete mra;
  mrb->collate(NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  mrb->reduce(&sum, NULL);
  MPI_Barrier(MPI_COMM_WORLD);
  mrb->gather(1);
  MPI_Barrier(MPI_COMM_WORLD);
  mrb->sort_keys(&ncompare);
  MPI_Barrier(MPI_COMM_WORLD);
  FILE * pFile4;
  pFile4 = fopen("hist.b", "w");
  mrb->map(mrb, &histoutput, pFile4);
  fclose(pFile4);
  delete mrb;
  MPI_Barrier(MPI_COMM_WORLD);
  pthread_cancel(thread1);
  MPI_Finalize();
//exit(0);
}