diff options
Diffstat (limited to 'icing/transform/icu/icu-normalizer_benchmark.cc')
| -rw-r--r-- | icing/transform/icu/icu-normalizer_benchmark.cc | 47 |
1 files changed, 39 insertions, 8 deletions
diff --git a/icing/transform/icu/icu-normalizer_benchmark.cc b/icing/transform/icu/icu-normalizer_benchmark.cc index fe8289a..89d5f1e 100644 --- a/icing/transform/icu/icu-normalizer_benchmark.cc +++ b/icing/transform/icu/icu-normalizer_benchmark.cc @@ -39,8 +39,8 @@ // blaze-bin/icing/transform/icu/icu-normalizer_benchmark // /data/local/tmp/ // -// $ adb shell /data/local/tmp/icu-normalizer_benchmark --benchmark_filter=all -// --adb +// $ adb shell /data/local/tmp/icu-normalizer_benchmark +// --benchmark_filter=all --adb // Flag to tell the benchmark that it'll be run on an Android device via adb, // the benchmark will set up data files accordingly. @@ -61,7 +61,6 @@ void BM_NormalizeUppercase(benchmark::State& state) { ICING_ASSERT_OK_AND_ASSIGN( std::unique_ptr<Normalizer> normalizer, normalizer_factory::Create( - /*max_term_byte_size=*/std::numeric_limits<int>::max())); std::string input_string(state.range(0), 'A'); @@ -95,7 +94,6 @@ void BM_NormalizeAccent(benchmark::State& state) { ICING_ASSERT_OK_AND_ASSIGN( std::unique_ptr<Normalizer> normalizer, normalizer_factory::Create( - /*max_term_byte_size=*/std::numeric_limits<int>::max())); std::string input_string; @@ -123,7 +121,7 @@ BENCHMARK(BM_NormalizeAccent) ->Arg(2048000) ->Arg(4096000); -void BM_NormalizeHiragana(benchmark::State& state) { +void BM_NormalizeGreekAccent(benchmark::State& state) { bool run_via_adb = absl::GetFlag(FLAGS_adb); if (!run_via_adb) { ICING_ASSERT_OK(icu_data_file_helper::SetUpICUDataFile( @@ -133,7 +131,43 @@ void BM_NormalizeHiragana(benchmark::State& state) { ICING_ASSERT_OK_AND_ASSIGN( std::unique_ptr<Normalizer> normalizer, normalizer_factory::Create( + /*max_term_byte_size=*/std::numeric_limits<int>::max())); + + std::string input_string; + while (input_string.length() < state.range(0)) { + input_string.append("άὰᾶἀἄ"); + } + + for (auto _ : state) { + normalizer->NormalizeTerm(input_string); + } +} +BENCHMARK(BM_NormalizeGreekAccent) + ->Arg(1000) + ->Arg(2000) + ->Arg(4000) + ->Arg(8000) + ->Arg(16000) + ->Arg(32000) + ->Arg(64000) + ->Arg(128000) + ->Arg(256000) + ->Arg(384000) + ->Arg(512000) + ->Arg(1024000) + ->Arg(2048000) + ->Arg(4096000); + +void BM_NormalizeHiragana(benchmark::State& state) { + bool run_via_adb = absl::GetFlag(FLAGS_adb); + if (!run_via_adb) { + ICING_ASSERT_OK(icu_data_file_helper::SetUpICUDataFile( + GetTestFilePath("icing/icu.dat"))); + } + ICING_ASSERT_OK_AND_ASSIGN( + std::unique_ptr<Normalizer> normalizer, + normalizer_factory::Create( /*max_term_byte_size=*/std::numeric_limits<int>::max())); std::string input_string; @@ -171,7 +205,6 @@ void BM_UppercaseSubTokenLength(benchmark::State& state) { ICING_ASSERT_OK_AND_ASSIGN( std::unique_ptr<Normalizer> normalizer, normalizer_factory::Create( - /*max_term_byte_size=*/std::numeric_limits<int>::max())); std::string input_string(state.range(0), 'A'); @@ -207,7 +240,6 @@ void BM_AccentSubTokenLength(benchmark::State& state) { ICING_ASSERT_OK_AND_ASSIGN( std::unique_ptr<Normalizer> normalizer, normalizer_factory::Create( - /*max_term_byte_size=*/std::numeric_limits<int>::max())); std::string input_string; @@ -248,7 +280,6 @@ void BM_HiraganaSubTokenLength(benchmark::State& state) { ICING_ASSERT_OK_AND_ASSIGN( std::unique_ptr<Normalizer> normalizer, normalizer_factory::Create( - /*max_term_byte_size=*/std::numeric_limits<int>::max())); std::string input_string; |