Function BM_ReadColumnIndex

cpp/src/parquet/page_index_benchmark.cc:66–100

template <typename DType>
void BM_ReadColumnIndex(::benchmark::State& state) {
  schema::NodePtr type = ::parquet::schema::PrimitiveNode::Make(
      "b", Repetition::OPTIONAL, DType::type_num, ConvertedType::NONE, 8);
  auto descr_ptr =
      std::make_unique<ColumnDescriptor>(type, /*def_level=*/1, /*rep_level=*/0);
  auto descr = descr_ptr.get();

  const int num_pages = static_cast<int>(state.range(0));
  auto builder = ColumnIndexBuilder::Make(descr);

  const size_t values_per_page = 100;
  for (int i = 0; i < num_pages; ++i) {
    auto stats = MakeStatistics<DType>(descr);
    std::vector<uint8_t> heap;
    std::vector<typename DType::c_type> values;
    values.resize(values_per_page);
    GenerateBenchmarkData(values_per_page, /*seed=*/0, values.data(), &heap,
                          kDataStringLength);
    stats->Update(values.data(), values_per_page, /*null_count=*/0);
    builder->AddPage(stats->Encode(), /*size_stats=*/{});
  }

  builder->Finish();
  auto sink = CreateOutputStream();
  builder->WriteTo(sink.get());
  auto buffer = sink->Finish().ValueOrDie();
  ReaderProperties properties;
  for (auto _ : state) {
    auto column_index = ColumnIndex::Make(*descr, buffer->data() + 0,
                                          static_cast<int>(buffer->size()), properties);
    ::benchmark::DoNotOptimize(column_index);
  }
  state.SetBytesProcessed(state.iterations() * buffer->size());
  state.SetItemsProcessed(state.iterations() * num_pages);
}

BENCHMARK_TEMPLATE(BM_ReadColumnIndex, Int64Type)->Apply(PageIndexSetArgs);
BENCHMARK_TEMPLATE(BM_ReadColumnIndex, DoubleType)->Apply(PageIndexSetArgs);