comparison query.cpp @ 539:06ed85832c3b multiprobeLSH

Optimized the query_loop_points inner loop for memcpy and I/O efficiency. Uses sparse seeks and reads to perform scattered reads across data set. Current version does not cache fid between open calls to the same trackID.

 
  error:
   return 1;
 }
 
-int audiodb_track_id_datum(adb_t *adb, uint32_t track_id, adb_datum_t *d) {
+int audiodb_track_id_datum(adb_t *adb, uint32_t track_id, adb_datum_t *d, off_t vector_offset=0, size_t num_vectors=0) {
   off_t track_offset = (*adb->track_offsets)[track_id];
   if(adb->header->flags & ADB_HEADER_FLAG_REFERENCES) {
     /* create a reference/insert, then use adb_insert_create_datum() */
     adb_reference_t reference = {0};
     char features[ADB_MAXSTR], power[ADB_MAXSTR], times[ADB_MAXSTR];

     if(adb->header->flags & ADB_HEADER_FLAG_TIMES) {
       lseek(adb->fd, adb->header->timesTableOffset + track_id * ADB_FILETABLE_ENTRY_SIZE, SEEK_SET);
       read_or_goto_error(adb->fd, times, ADB_MAXSTR);
       reference.times = times;
     }
-    return audiodb_insert_create_datum(&reference, d);
+    return audiodb_insert_create_datum(&reference, d, vector_offset*adb->header->dim*sizeof(double), num_vectors*adb->header->dim*sizeof(double));
   } else {
     /* initialize from sources of data that we already have */
-    d->nvectors = (*adb->track_lengths)[track_id];
+    if(num_vectors)
+      d->nvectors = num_vectors;
+    else
+      d->nvectors = (*adb->track_lengths)[track_id];
     d->dim = adb->header->dim;
     d->key = (*adb->keys)[track_id].c_str();
     /* read out stuff from the database tables */
     d->data = (double *) malloc(d->nvectors * d->dim * sizeof(double));
-    lseek(adb->fd, adb->header->dataOffset + track_offset, SEEK_SET);
+    lseek(adb->fd, adb->header->dataOffset + track_offset + vector_offset*d->dim*sizeof(double), SEEK_SET);
     read_or_goto_error(adb->fd, d->data, d->nvectors * d->dim * sizeof(double));
     if(adb->header->flags & ADB_HEADER_FLAG_POWER) {
       d->power = (double *) malloc(d->nvectors * sizeof(double));
-      lseek(adb->fd, adb->header->powerTableOffset + track_offset / d->dim, SEEK_SET);
+      lseek(adb->fd, adb->header->powerTableOffset + track_offset / d->dim + vector_offset*sizeof(double), SEEK_SET);
       read_or_goto_error(adb->fd, d->power, d->nvectors * sizeof(double));
     }
     if(adb->header->flags & ADB_HEADER_FLAG_TIMES) {
       d->times = (double *) malloc(2 * d->nvectors * sizeof(double));
-      lseek(adb->fd, adb->header->timesTableOffset + track_offset / d->dim, SEEK_SET);
+      lseek(adb->fd, adb->header->timesTableOffset + track_offset / d->dim + 2 * vector_offset*sizeof(double), SEEK_SET);
       read_or_goto_error(adb->fd, d->times, 2 * d->nvectors * sizeof(double));
     }
     return 0;
   }
  error:

 int audiodb_datum_qpointers_partial(adb_datum_t *d, uint32_t sequence_length, double **vector_data, 
 				    double **vector, adb_qpointers_internal_t *qpointers, 
 				    adb_qstate_internal_t *qstate){
   uint32_t nvectors = d->nvectors;
   qpointers->nvectors = nvectors;
-  std::priority_queue<PointPair, std::vector<PointPair>, greater<PointPair> > ppairs(*qstate->exact_evaluation_queue);
-  
-  size_t vector_size = nvectors * sizeof(double) * d->dim;
-
-  if(d->power)
-    qpointers->power_data = new double[vector_size / d->dim];
-
-  uint32_t seq_len_dbl = sequence_length*sizeof(double);
-  PointPair pp = ppairs.top();
-  uint32_t tid = pp.trackID;
-
-  while( !ppairs.empty() && pp.trackID==tid){
-    uint32_t spos = pp.spos;
+  
+  PointPair pp = (*qstate->exact_evaluation_queue).top();
 #ifdef _LSH_DEBUG_
-    cout << "tid=" << pp.trackID << " qpos=" << pp.qpos << " spos=" << pp.spos << endl;
-    cout.flush();
+  cout << "tid=" << pp.trackID << " qpos=" << pp.qpos << " spos=" << pp.spos << endl;
+  cout.flush();
 #endif
-
-    if(d->power) {
-      memcpy(qpointers->power_data+spos, d->power+spos, seq_len_dbl);
-      audiodb_sequence_sum(qpointers->power_data+spos, sequence_length, sequence_length);
-      audiodb_sequence_average(qpointers->power_data+spos, sequence_length, sequence_length);
-    }
-    ppairs.pop();
-    if(!ppairs.empty())
-      pp = ppairs.top();
-  }
-
+  
+  if(d->power) {
+    //memcpy(qpointers->power_data, d->power, seq_len_dbl);
+    audiodb_sequence_sum(d->power, sequence_length, sequence_length);
+    audiodb_sequence_average(d->power, sequence_length, sequence_length);
+  }
+  
   if(d->times) {
     qpointers->mean_duration = new double[1];
     *qpointers->mean_duration = 0;
     for(unsigned int k = 0; k < nvectors; k++) {
       *qpointers->mean_duration += d->times[2*k+1] - d->times[2*k];
     }
     *qpointers->mean_duration /= nvectors;
   }  
-
+  
   *vector = d->data;
   *vector_data = d->data;
   qpointers->l2norm = 0 ;
-  qpointers->power = qpointers->power_data;
+  qpointers->power = d->power;
   return 0;
 }
 
 int audiodb_query_spec_qpointers(adb_t *adb, const adb_query_spec_t *spec, double **vector_data, double **vector, adb_qpointers_internal_t *qpointers) {
   adb_datum_t *datum;

    * don't overlap, so we will use exhaustive dot product evaluation
    * (instead of memoization of partial sums, as in query_loop()).
    */
   double dist;
   double *dbdata = 0, *dbdata_pointer;
-  Uns32T currentTrack = 0x80000000; // KLUDGE: Initialize with a value outside of track index range
   Uns32T npairs = qstate->exact_evaluation_queue->size();
 #ifdef _LSH_DEBUG_
   cout << "Num vector pairs to evaluate: " << npairs << "..." << endl;
   cout.flush();
 #endif  
   adb_datum_t d = {0};
   while(npairs--) {
     PointPair pp = qstate->exact_evaluation_queue->top();
-    if(currentTrack != pp.trackID) {
-      maybe_delete_array(dbpointers.power_data);
-      maybe_delete_array(dbpointers.mean_duration);
-      currentTrack = pp.trackID;
+    maybe_delete_array(dbpointers.mean_duration);
+    if(audiodb_track_id_datum(adb, pp.trackID, &d, pp.spos, sequence_length)) {
+      delete qstate->exact_evaluation_queue;
+      delete qstate->set;
+      return 1;
+    }
+      
+    if(audiodb_datum_qpointers_partial(&d, sequence_length, &dbdata, &dbdata_pointer, &dbpointers, qstate)) {
+      delete qstate->exact_evaluation_queue;
+      delete qstate->set;
       audiodb_free_datum(&d);
-      if(audiodb_track_id_datum(adb, pp.trackID, &d)) {
-        delete qstate->exact_evaluation_queue;
-        delete qstate->set;
-        return 1;
-      }
+      return 1;
+    }
       
-      if(audiodb_datum_qpointers_partial(&d, sequence_length, &dbdata, &dbdata_pointer, &dbpointers, qstate)) {
-        delete qstate->exact_evaluation_queue;
-        delete qstate->set;
-        audiodb_free_datum(&d);
-        return 1;
-      }      
-    }
     Uns32T qPos = (spec->qid.flags & ADB_QID_FLAG_EXHAUSTIVE) ? pp.qpos : 0;
-    Uns32T sPos = pp.spos; // index into l2norm table
     // Test power thresholds before computing distance
-    if( ( (!power_refine) || audiodb_powers_acceptable(&spec->refine, qpointers->power[qPos], dbpointers.power[sPos])) &&
-	( qPos<qpointers->nvectors-sequence_length+1 && sPos<(*adb->track_lengths)[pp.trackID]-sequence_length+1 ) ){
+    if( ( (!power_refine) || audiodb_powers_acceptable(&spec->refine, qpointers->power[qPos], dbpointers.power[0])) &&
+	( qPos<qpointers->nvectors-sequence_length+1 && pp.spos<(*adb->track_lengths)[pp.trackID]-sequence_length+1 ) ){
       // Compute distance    
-      dist = audiodb_dot_product(query + qPos*adb->header->dim, dbdata + sPos*adb->header->dim, adb->header->dim*sequence_length);
+      dist = audiodb_dot_product(query + qPos*adb->header->dim, dbdata, adb->header->dim*sequence_length);
       double qn = audiodb_dot_product(query + qPos*adb->header->dim, query + qPos*adb->header->dim, adb->header->dim*sequence_length);
-      double sn = audiodb_dot_product(dbdata + sPos*adb->header->dim, dbdata + sPos*adb->header->dim, adb->header->dim*sequence_length);
+      double sn = audiodb_dot_product(dbdata, dbdata, adb->header->dim*sequence_length);
       qn = sqrt(qn);
       sn = sqrt(sn);
       switch(spec->params.distance) {
       case ADB_DISTANCE_EUCLIDEAN_NORMED:
 	dist = 2 - (2/(qn*sn))*dist;

         r.ipos = pp.spos;
         qstate->accumulator->add_point(&r);
       }
     }
     qstate->exact_evaluation_queue->pop();
+    audiodb_free_datum(&d);
   }
 
   // Cleanup
-  audiodb_free_datum(&d);
   //  maybe_delete_array(dbdata);
   //maybe_delete_array(dbpointers.l2norm_data);
-  maybe_delete_array(dbpointers.power_data);
+  //maybe_delete_array(dbpointers.power_data);
   maybe_delete_array(dbpointers.mean_duration);
   delete qstate->exact_evaluation_queue;
   delete qstate->set;
   return 0;
 }