Use east const in benchmarks files (#13491)

Use east const in _cpp/benchmarks/_ files
used clang-format `QualifierAlignment: Right` to do the clean up.

since https://clang.llvm.org/docs/ClangFormatStyleOptions.html#qualifieralignment has warning that 
`Setting QualifierAlignment to something other than Leave, COULD lead to incorrect code formatting due to incorrect decisions made due to clang-formats lack of complete semantic information. As such extra care should be taken to review code changes made by the use of this option.`
So it's not added to .clang-format

Split to multiple PRs for easy reviewing #13491, #13492, #13493, #13494

Authors:
  - Karthikeyan (https://github.com/karthikeyann)
  - Nghia Truong (https://github.com/ttnghia)

Approvers:
  - Nghia Truong (https://github.com/ttnghia)
  - Bradley Dice (https://github.com/bdice)

URL: #13491

cpp/benchmarks/copying/contiguous_split.cu

@@ -126,7 +126,7 @@ void BM_contiguous_split_strings(benchmark::State& state, ContiguousSplitImpl& i
   bool const include_validity       = state.range(3) != 0;
 
   constexpr int64_t string_len = 8;
-  std::vector<const char*> h_strings{
+  std::vector<char const*> h_strings{
     "aaaaaaaa", "bbbbbbbb", "cccccccc", "dddddddd", "eeeeeeee", "ffffffff", "gggggggg", "hhhhhhhh"};
 
   int64_t const col_len_bytes = total_desired_bytes / num_cols;

cpp/benchmarks/copying/gather.cu

@@ -31,8 +31,8 @@ class Gather : public cudf::benchmark {};
 template <class TypeParam, bool coalesce>
 void BM_gather(benchmark::State& state)
 {
-  const cudf::size_type source_size{(cudf::size_type)state.range(0)};
-  const auto n_cols = (cudf::size_type)state.range(1);
+  cudf::size_type const source_size{(cudf::size_type)state.range(0)};
+  auto const n_cols = (cudf::size_type)state.range(1);
 
   // Gather indices
   auto gather_map_table =

cpp/benchmarks/fixture/benchmark_fixture.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -73,28 +73,28 @@ class benchmark : public ::benchmark::Fixture {
  public:
   benchmark() : ::benchmark::Fixture()
   {
-    const char* env_iterations = std::getenv("CUDF_BENCHMARK_ITERATIONS");
+    char const* env_iterations = std::getenv("CUDF_BENCHMARK_ITERATIONS");
     if (env_iterations != nullptr) { this->Iterations(std::max(0L, atol(env_iterations))); }
   }
 
-  void SetUp(const ::benchmark::State& state) override
+  void SetUp(::benchmark::State const& state) override
   {
     mr = make_pool_instance();
     rmm::mr::set_current_device_resource(mr.get());  // set default resource to pool
   }
 
-  void TearDown(const ::benchmark::State& state) override
+  void TearDown(::benchmark::State const& state) override
   {
     // reset default resource to the initial resource
     rmm::mr::set_current_device_resource(nullptr);
     mr.reset();
   }
 
   // eliminate partial override warnings (see benchmark/benchmark.h)
-  void SetUp(::benchmark::State& st) override { SetUp(const_cast<const ::benchmark::State&>(st)); }
+  void SetUp(::benchmark::State& st) override { SetUp(const_cast<::benchmark::State const&>(st)); }
   void TearDown(::benchmark::State& st) override
   {
-    TearDown(const_cast<const ::benchmark::State&>(st));
+    TearDown(const_cast<::benchmark::State const&>(st));
   }
 
   std::shared_ptr<rmm::mr::device_memory_resource> mr;

cpp/benchmarks/fixture/templated_benchmark_fixture.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021, NVIDIA CORPORATION.
+ * Copyright (c) 2021-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -45,7 +45,7 @@ namespace cudf {
 template <class Fixture>
 class FunctionTemplateBenchmark : public Fixture {
  public:
-  FunctionTemplateBenchmark(const char* name, ::benchmark::internal::Function* func)
+  FunctionTemplateBenchmark(char const* name, ::benchmark::internal::Function* func)
     : Fixture(), func_(func)
   {
     this->SetName(name);

cpp/benchmarks/groupby/group_max.cpp

@@ -23,7 +23,7 @@
 template <typename Type>
 void bench_groupby_max(nvbench::state& state, nvbench::type_list<Type>)
 {
-  const auto size = static_cast<cudf::size_type>(state.get_int64("num_rows"));
+  auto const size = static_cast<cudf::size_type>(state.get_int64("num_rows"));
 
   auto const keys = [&] {
     data_profile const profile = data_profile_builder().cardinality(0).no_validity().distribution(

cpp/benchmarks/groupby/group_no_requests.cpp

@@ -28,7 +28,7 @@ class Groupby : public cudf::benchmark {};
 
 void BM_basic_no_requests(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
 
   data_profile const profile = data_profile_builder().cardinality(0).no_validity().distribution(
     cudf::type_to_id<int64_t>(), distribution_id::UNIFORM, 0, 100);
@@ -59,7 +59,7 @@ BENCHMARK_REGISTER_F(Groupby, BasicNoRequest)
 
 void BM_pre_sorted_no_requests(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
 
   data_profile const profile = data_profile_builder().cardinality(0).no_validity().distribution(
     cudf::type_to_id<int64_t>(), distribution_id::UNIFORM, 0, 100);

cpp/benchmarks/groupby/group_nth.cpp

@@ -29,7 +29,7 @@ class Groupby : public cudf::benchmark {};
 void BM_pre_sorted_nth(benchmark::State& state)
 {
   // const cudf::size_type num_columns{(cudf::size_type)state.range(0)};
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
 
   data_profile const profile = data_profile_builder().cardinality(0).no_validity().distribution(
     cudf::type_to_id<int64_t>(), distribution_id::UNIFORM, 0, 100);

cpp/benchmarks/groupby/group_nunique.cpp

@@ -39,7 +39,7 @@ auto make_aggregation_request_vector(cudf::column_view const& values, Args&&...
 template <typename Type>
 void bench_groupby_nunique(nvbench::state& state, nvbench::type_list<Type>)
 {
-  const auto size = static_cast<cudf::size_type>(state.get_int64("num_rows"));
+  auto const size = static_cast<cudf::size_type>(state.get_int64("num_rows"));
 
   auto const keys = [&] {
     data_profile profile = data_profile_builder().cardinality(0).no_validity().distribution(

cpp/benchmarks/groupby/group_scan.cpp

@@ -29,7 +29,7 @@ class Groupby : public cudf::benchmark {};
 
 void BM_basic_sum_scan(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
 
   data_profile const profile = data_profile_builder().cardinality(0).no_validity().distribution(
     cudf::type_to_id<int64_t>(), distribution_id::UNIFORM, 0, 100);
@@ -61,7 +61,7 @@ BENCHMARK_REGISTER_F(Groupby, BasicSumScan)
 
 void BM_pre_sorted_sum_scan(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
 
   data_profile profile = data_profile_builder().cardinality(0).no_validity().distribution(
     cudf::type_to_id<int64_t>(), distribution_id::UNIFORM, 0, 100);

cpp/benchmarks/groupby/group_shift.cpp

@@ -28,8 +28,8 @@ class Groupby : public cudf::benchmark {};
 
 void BM_group_shift(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
-  const int num_groups = 100;
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
+  int const num_groups = 100;
 
   data_profile const profile =
     data_profile_builder().cardinality(0).null_probability(0.01).distribution(

cpp/benchmarks/groupby/group_struct_keys.cpp

@@ -33,10 +33,10 @@ void bench_groupby_struct_keys(nvbench::state& state)
   std::default_random_engine generator;
   std::uniform_int_distribution<int> distribution(0, 100);
 
-  const cudf::size_type n_rows{static_cast<cudf::size_type>(state.get_int64("NumRows"))};
-  const cudf::size_type n_cols{1};
-  const cudf::size_type depth{static_cast<cudf::size_type>(state.get_int64("Depth"))};
-  const bool nulls{static_cast<bool>(state.get_int64("Nulls"))};
+  cudf::size_type const n_rows{static_cast<cudf::size_type>(state.get_int64("NumRows"))};
+  cudf::size_type const n_cols{1};
+  cudf::size_type const depth{static_cast<cudf::size_type>(state.get_int64("Depth"))};
+  bool const nulls{static_cast<bool>(state.get_int64("Nulls"))};
 
   // Create columns with values in the range [0,100)
   std::vector<column_wrapper> columns;

cpp/benchmarks/groupby/group_sum.cpp

@@ -28,7 +28,7 @@ class Groupby : public cudf::benchmark {};
 
 void BM_basic_sum(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
 
   data_profile const profile = data_profile_builder().cardinality(0).no_validity().distribution(
     cudf::type_to_id<int64_t>(), distribution_id::UNIFORM, 0, 100);
@@ -61,7 +61,7 @@ BENCHMARK_REGISTER_F(Groupby, Basic)
 
 void BM_pre_sorted_sum(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
 
   data_profile profile = data_profile_builder().cardinality(0).no_validity().distribution(
     cudf::type_to_id<int64_t>(), distribution_id::UNIFORM, 0, 100);

cpp/benchmarks/io/fst.cu

@@ -60,7 +60,7 @@ auto make_test_json_data(nvbench::state& state)
 
   auto d_input_scalar                = cudf::make_string_scalar(input);
   auto& d_string_scalar              = static_cast<cudf::string_scalar&>(*d_input_scalar);
-  const cudf::size_type repeat_times = string_size / input.size();
+  cudf::size_type const repeat_times = string_size / input.size();
   return cudf::strings::repeat_string(d_string_scalar, repeat_times);
 }
 

cpp/benchmarks/io/json/nested_json.cpp

@@ -77,7 +77,7 @@ std::string generate_row(
   int num_columns, int max_depth, int max_list_size, int max_struct_size, size_t max_bytes)
 {
   std::string s = "{";
-  const std::vector<std::string> elems{
+  std::vector<std::string> const elems{
     R"(1)", R"(-2)", R"(3.4)", R"("5")", R"("abcdefghij")", R"(true)", R"(null)"};
   for (int i = 0; i < num_columns; i++) {
     s += R"("col)" + num_to_string(i) + R"(": )";
@@ -140,7 +140,7 @@ auto make_test_json_data(cudf::size_type string_size, rmm::cuda_stream_view stre
                       {"a":1,"b":Infinity,"c":[null], "d": {"year":-600,"author": "Kaniyan"}},
                       {"a": 1, "b": 8.0, "d": { "author": "Jean-Jacques Rousseau"}},)";
 
-  const cudf::size_type repeat_times = string_size / input.size();
+  cudf::size_type const repeat_times = string_size / input.size();
 
   auto d_input_scalar   = cudf::make_string_scalar(input, stream);
   auto& d_string_scalar = static_cast<cudf::string_scalar&>(*d_input_scalar);
@@ -191,7 +191,7 @@ void BM_NESTED_JSON_DEPTH(nvbench::state& state)
 
   auto d_scalar = cudf::string_scalar(
     generate_json(100'000'000, 10, depth, 10, 10, string_size), true, cudf::get_default_stream());
-  auto input = cudf::device_span<const char>(d_scalar.data(), d_scalar.size());
+  auto input = cudf::device_span<char const>(d_scalar.data(), d_scalar.size());
 
   state.add_element_count(input.size());
   auto const default_options = cudf::io::json_reader_options{};

cpp/benchmarks/iterator/iterator.cu

@@ -131,7 +131,7 @@ class Iterator : public cudf::benchmark {};
 template <class TypeParam, bool cub_or_thrust, bool raw_or_iterator>
 void BM_iterator(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
   using T      = TypeParam;
   auto num_gen = thrust::counting_iterator<cudf::size_type>(0);
 
@@ -195,7 +195,7 @@ void pair_iterator_bench_thrust(cudf::column_view& col,
 template <class TypeParam, bool cub_or_thrust>
 void BM_pair_iterator(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
   using T      = TypeParam;
   auto num_gen = thrust::counting_iterator<cudf::size_type>(0);
   auto null_gen =

cpp/benchmarks/join/generate_input_tables.cuh

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2023, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -31,7 +31,7 @@
 
 #include <cassert>
 
-__global__ static void init_curand(curandState* state, const int nstates)
+__global__ static void init_curand(curandState* state, int const nstates)
 {
   int ithread = threadIdx.x + blockIdx.x * blockDim.x;
 
@@ -40,10 +40,10 @@ __global__ static void init_curand(curandState* state, const int nstates)
 
 template <typename key_type, typename size_type>
 __global__ static void init_build_tbl(key_type* const build_tbl,
-                                      const size_type build_tbl_size,
-                                      const int multiplicity,
+                                      size_type const build_tbl_size,
+                                      int const multiplicity,
                                       curandState* state,
-                                      const int num_states)
+                                      int const num_states)
 {
   auto const start_idx = blockIdx.x * blockDim.x + threadIdx.x;
   auto const stride    = blockDim.x * gridDim.x;
@@ -52,7 +52,7 @@ __global__ static void init_build_tbl(key_type* const build_tbl,
   curandState localState = state[start_idx];
 
   for (size_type idx = start_idx; idx < build_tbl_size; idx += stride) {
-    const double x = curand_uniform_double(&localState);
+    double const x = curand_uniform_double(&localState);
 
     build_tbl[idx] = static_cast<key_type>(x * (build_tbl_size / multiplicity));
   }
@@ -62,13 +62,13 @@ __global__ static void init_build_tbl(key_type* const build_tbl,
 
 template <typename key_type, typename size_type>
 __global__ void init_probe_tbl(key_type* const probe_tbl,
-                               const size_type probe_tbl_size,
-                               const size_type build_tbl_size,
-                               const key_type rand_max,
-                               const double selectivity,
-                               const int multiplicity,
+                               size_type const probe_tbl_size,
+                               size_type const build_tbl_size,
+                               key_type const rand_max,
+                               double const selectivity,
+                               int const multiplicity,
                                curandState* state,
-                               const int num_states)
+                               int const num_states)
 {
   auto const start_idx = blockIdx.x * blockDim.x + threadIdx.x;
   auto const stride    = blockDim.x * gridDim.x;
@@ -123,11 +123,11 @@ __global__ void init_probe_tbl(key_type* const probe_tbl,
  */
 template <typename key_type, typename size_type>
 void generate_input_tables(key_type* const build_tbl,
-                           const size_type build_tbl_size,
+                           size_type const build_tbl_size,
                            key_type* const probe_tbl,
-                           const size_type probe_tbl_size,
-                           const double selectivity,
-                           const int multiplicity)
+                           size_type const probe_tbl_size,
+                           double const selectivity,
+                           int const multiplicity)
 {
   // With large values of rand_max the a lot of temporary storage is needed for the lottery. At the
   // expense of not being that accurate with applying the selectivity an especially more memory
@@ -152,7 +152,7 @@ void generate_input_tables(key_type* const build_tbl,
   int num_sms{-1};
   CUDF_CUDA_TRY(cudaDeviceGetAttribute(&num_sms, cudaDevAttrMultiProcessorCount, dev_id));
 
-  const int num_states =
+  int const num_states =
     num_sms * std::max(num_blocks_init_build_tbl, num_blocks_init_probe_tbl) * block_size;
   rmm::device_uvector<curandState> devStates(num_states, cudf::get_default_stream());
 

cpp/benchmarks/join/join_common.hpp

@@ -96,8 +96,8 @@ void BM_join(state_type& state, Join JoinFunc)
     }
   }();
 
-  const double selectivity = 0.3;
-  const int multiplicity   = 1;
+  double const selectivity = 0.3;
+  int const multiplicity   = 1;
 
   // Generate build and probe tables
   auto build_random_null_mask = [](int size) {

cpp/benchmarks/lists/copying/scatter_lists.cu

@@ -40,9 +40,9 @@ void BM_lists_scatter(::benchmark::State& state)
   auto stream = cudf::get_default_stream();
   auto mr     = rmm::mr::get_current_device_resource();
 
-  const cudf::size_type base_size{(cudf::size_type)state.range(0)};
-  const cudf::size_type num_elements_per_row{(cudf::size_type)state.range(1)};
-  const auto num_rows = (cudf::size_type)ceil(double(base_size) / num_elements_per_row);
+  cudf::size_type const base_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const num_elements_per_row{(cudf::size_type)state.range(1)};
+  auto const num_rows = (cudf::size_type)ceil(double(base_size) / num_elements_per_row);
 
   auto source_base_col = make_fixed_width_column(cudf::data_type{cudf::type_to_id<TypeParam>()},
                                                  base_size,

cpp/benchmarks/null_mask/set_null_mask.cpp

@@ -23,7 +23,7 @@ class SetNullmask : public cudf::benchmark {};
 
 void BM_setnullmask(benchmark::State& state)
 {
-  const cudf::size_type size{(cudf::size_type)state.range(0)};
+  cudf::size_type const size{(cudf::size_type)state.range(0)};
   rmm::device_buffer mask = cudf::create_null_mask(size, cudf::mask_state::UNINITIALIZED);
   auto begin = 0, end = size;
 

cpp/benchmarks/quantiles/quantiles.cpp

@@ -30,9 +30,9 @@ static void BM_quantiles(benchmark::State& state, bool nulls)
 {
   using Type = int;
 
-  const cudf::size_type n_rows{(cudf::size_type)state.range(0)};
-  const cudf::size_type n_cols{(cudf::size_type)state.range(1)};
-  const cudf::size_type n_quantiles{(cudf::size_type)state.range(2)};
+  cudf::size_type const n_rows{(cudf::size_type)state.range(0)};
+  cudf::size_type const n_cols{(cudf::size_type)state.range(1)};
+  cudf::size_type const n_quantiles{(cudf::size_type)state.range(2)};
 
   // Create columns with values in the range [0,100)
   data_profile profile = data_profile_builder().cardinality(0).distribution(

cpp/benchmarks/reduction/anyall.cpp

@@ -30,7 +30,7 @@ template <typename type>
 void BM_reduction_anyall(benchmark::State& state,
                          std::unique_ptr<cudf::reduce_aggregation> const& agg)
 {
-  const cudf::size_type column_size{static_cast<cudf::size_type>(state.range(0))};
+  cudf::size_type const column_size{static_cast<cudf::size_type>(state.range(0))};
   auto const dtype           = cudf::type_to_id<type>();
   data_profile const profile = data_profile_builder().no_validity().distribution(
     dtype, distribution_id::UNIFORM, 0, agg->kind == cudf::aggregation::ANY ? 0 : 100);

cpp/benchmarks/reduction/dictionary.cpp

@@ -29,7 +29,7 @@ template <typename T>
 void BM_reduction_dictionary(benchmark::State& state,
                              std::unique_ptr<cudf::reduce_aggregation> const& agg)
 {
-  const cudf::size_type column_size{static_cast<cudf::size_type>(state.range(0))};
+  cudf::size_type const column_size{static_cast<cudf::size_type>(state.range(0))};
 
   // int column and encoded dictionary column
   data_profile const profile = data_profile_builder().cardinality(0).no_validity().distribution(

cpp/benchmarks/reduction/minmax.cpp

@@ -27,7 +27,7 @@ class Reduction : public cudf::benchmark {};
 template <typename type>
 void BM_reduction(benchmark::State& state)
 {
-  const cudf::size_type column_size{(cudf::size_type)state.range(0)};
+  cudf::size_type const column_size{(cudf::size_type)state.range(0)};
   auto const dtype = cudf::type_to_id<type>();
   auto const input_column =
     create_random_column(dtype, row_count{column_size}, data_profile_builder().no_validity());