Method fixed_sample_size_run

libCacheSim/mrcProfiler/mrcProfiler.cpp:171–265

}

void mrcProfiler::MRCProfilerSHARDS::fixed_sample_size_run() {
  // 1. init
  request_t *req = new_request();
  double sample_rate = 1.0;
  std::vector<double> local_hit_cnt_vec(mrc_size_vec.size(), 0);
  std::vector<double> local_hit_size_vec(mrc_size_vec.size(), 0);
  double sampled_cnt = 0, sampled_size = 0;
  int64_t current_time = 0;
  int64_t max_to_keep = params_.shards_params.sample_size;

  MinValueMap<int64_t, uint64_t> min_value_map(max_to_keep);
  robin_hood::unordered_map<obj_id_t, int64_t> last_access_time_map;
  SplayTree<int64_t, uint64_t> rd_tree;

  // 2. go through the trace
  read_one_req(reader_, req);
  /* going through the trace */
  do {
    DEBUG_ASSERT(req->obj_size != 0);
    n_req_ += 1;
    sum_obj_size_req += req->obj_size;

    uint64_t hash_value = get_hash_value_int_64_with_salt(
        req->obj_id, params_.shards_params.salt);

    current_time += 1;
    if (!min_value_map.full() || hash_value < min_value_map.get_max_value() ||
        last_access_time_map.count(req->obj_id)) {
      // this is a sampled req

      if (!last_access_time_map.count(req->obj_id)) {
        bool poped = false;
        int64_t poped_id = min_value_map.insert(req->obj_id, hash_value, poped);
        if (poped) {
          // this is a sampled req
          int64_t poped_id_access_time = last_access_time_map[poped_id];
          rd_tree.erase(poped_id_access_time);
          last_access_time_map.erase(poped_id);
        }
      }

      if (!min_value_map.full()) {
        sample_rate = 1.0;  // still 100% sample rate
      } else {
        sample_rate = min_value_map.get_max_value() * 1.0 /
                      UINT64_MAX;  // adjust the sample rate
      }

      sampled_cnt += 1.0 / sample_rate;
      sampled_size += 1.0 * req->obj_size / sample_rate;

      if (last_access_time_map.count(req->obj_id)) {
        int64_t last_acc_time = last_access_time_map[req->obj_id];
        int64_t stack_distance =
            rd_tree.getDistance(last_acc_time) * 1.0 / sample_rate;

        last_access_time_map[req->obj_id] = current_time;