Move tpch examples into benchmarks

rapidsai · Aug 26, 2024 · d2d01d6 · d2d01d6
1 parent f511322
commit d2d01d6
Show file tree

Hide file tree

Showing 13 changed files with 9 additions and 63 deletions.
diff --git a/cpp/benchmarks/CMakeLists.txt b/cpp/benchmarks/CMakeLists.txt
@@ -155,6 +155,14 @@ ConfigureBench(TRANSPOSE_BENCH transpose/transpose.cpp)
 # * apply_boolean_mask benchmark ------------------------------------------------------------------
 ConfigureBench(APPLY_BOOLEAN_MASK_BENCH stream_compaction/apply_boolean_mask.cpp)
 
+# ##################################################################################################
+# * tpch benchmark --------------------------------------------------------------------------------
+ConfigureBench(TPCH_Q1 tpch/q1.cpp)
+ConfigureBench(TPCH_Q5 tpch/q5.cpp)
+ConfigureBench(TPCH_Q6 tpch/q6.cpp)
+ConfigureBench(TPCH_Q9 tpch/q9.cpp)
+ConfigureBench(TPCH_Q10 tpch/q10.cpp)
+
 # ##################################################################################################
 # * stream_compaction benchmark -------------------------------------------------------------------
 ConfigureNVBench(

diff --git a/cpp/examples/tpch/README.md → cpp/benchmarks/tpch/README.md b/cpp/examples/tpch/README.md → cpp/benchmarks/tpch/README.md
diff --git a/...xamples/tpch/datagen/correct_datatypes.py → ...chmarks/tpch/datagen/correct_datatypes.py b/...xamples/tpch/datagen/correct_datatypes.py → ...chmarks/tpch/datagen/correct_datatypes.py
diff --git a/cpp/examples/tpch/datagen/datagen.sh → cpp/benchmarks/tpch/datagen/datagen.sh b/cpp/examples/tpch/datagen/datagen.sh → cpp/benchmarks/tpch/datagen/datagen.sh
diff --git a/cpp/examples/tpch/datagen/tpch.patch → cpp/benchmarks/tpch/datagen/tpch.patch b/cpp/examples/tpch/datagen/tpch.patch → cpp/benchmarks/tpch/datagen/tpch.patch
diff --git a/cpp/examples/tpch/q1.cpp → cpp/benchmarks/tpch/q1.cpp b/cpp/examples/tpch/q1.cpp → cpp/benchmarks/tpch/q1.cpp
@@ -14,7 +14,6 @@
  * limitations under the License.
  */
 
-#include "../utilities/timer.hpp"
 #include "utils.hpp"
 
 #include <cudf/ast/expressions.hpp>
@@ -108,8 +107,6 @@ int main(int argc, char const** argv)
   auto resource = create_memory_resource(args.memory_resource_type);
   rmm::mr::set_current_device_resource(resource.get());
 
-  cudf::examples::timer timer;
-
   // Define the column projections and filter predicate for `lineitem` table
   std::vector<std::string> const lineitem_cols = {"l_returnflag",
                                                   "l_linestatus",
@@ -166,8 +163,6 @@ int main(int argc, char const** argv)
                                              {"l_returnflag", "l_linestatus"},
                                              {cudf::order::ASCENDING, cudf::order::ASCENDING});
 
-  timer.print_elapsed_millis();
-
   // Write query result to a parquet file
   orderedby_table->to_parquet("q1.parquet");
   return 0;

diff --git a/cpp/examples/tpch/q10.cpp → cpp/benchmarks/tpch/q10.cpp b/cpp/examples/tpch/q10.cpp → cpp/benchmarks/tpch/q10.cpp
@@ -14,7 +14,6 @@
  * limitations under the License.
  */
 
-#include "../utilities/timer.hpp"
 #include "utils.hpp"
 
 #include <cudf/ast/expressions.hpp>
@@ -97,8 +96,6 @@ int main(int argc, char const** argv)
   auto resource = create_memory_resource(args.memory_resource_type);
   rmm::mr::set_current_device_resource(resource.get());
 
-  cudf::examples::timer timer;
-
   // Define the column projection and filter predicate for the `orders` table
   std::vector<std::string> const orders_cols = {"o_custkey", "o_orderkey", "o_orderdate"};
   auto const o_orderdate_ref                 = cudf::ast::column_reference(std::distance(
@@ -158,8 +155,6 @@ int main(int argc, char const** argv)
   auto const orderedby_table =
     apply_orderby(groupedby_table, {"revenue"}, {cudf::order::DESCENDING});
 
-  timer.print_elapsed_millis();
-
   // Write query result to a parquet file
   orderedby_table->to_parquet("q10.parquet");
   return 0;

diff --git a/cpp/examples/tpch/q5.cpp → cpp/benchmarks/tpch/q5.cpp b/cpp/examples/tpch/q5.cpp → cpp/benchmarks/tpch/q5.cpp
@@ -14,7 +14,6 @@
  * limitations under the License.
  */
 
-#include "../utilities/timer.hpp"
 #include "utils.hpp"
 
 #include <cudf/ast/expressions.hpp>
@@ -93,8 +92,6 @@ int main(int argc, char const** argv)
   auto resource = create_memory_resource(args.memory_resource_type);
   rmm::mr::set_current_device_resource(resource.get());
 
-  cudf::examples::timer timer;
-
   // Define the column projection and filter predicate for the `orders` table
   std::vector<std::string> const orders_cols = {"o_custkey", "o_orderkey", "o_orderdate"};
   auto const o_orderdate_ref                 = cudf::ast::column_reference(std::distance(
@@ -161,8 +158,6 @@ int main(int argc, char const** argv)
   auto const orderedby_table =
     apply_orderby(groupedby_table, {"revenue"}, {cudf::order::DESCENDING});
 
-  timer.print_elapsed_millis();
-
   // Write query result to a parquet file
   orderedby_table->to_parquet("q5.parquet");
   return 0;

diff --git a/cpp/examples/tpch/q6.cpp → cpp/benchmarks/tpch/q6.cpp b/cpp/examples/tpch/q6.cpp → cpp/benchmarks/tpch/q6.cpp
@@ -14,7 +14,6 @@
  * limitations under the License.
  */
 
-#include "../utilities/timer.hpp"
 #include "utils.hpp"
 
 #include <cudf/ast/expressions.hpp>
@@ -67,8 +66,6 @@ int main(int argc, char const** argv)
   auto resource = create_memory_resource(args.memory_resource_type);
   rmm::mr::set_current_device_resource(resource.get());
 
-  cudf::examples::timer timer;
-
   // Read out the `lineitem` table from parquet file
   std::vector<std::string> const lineitem_cols = {
     "l_extendedprice", "l_discount", "l_shipdate", "l_quantity"};
@@ -129,8 +126,6 @@ int main(int argc, char const** argv)
   auto const revenue_view = revenue->view();
   auto const result_table = apply_reduction(revenue_view, cudf::aggregation::Kind::SUM, "revenue");
 
-  timer.print_elapsed_millis();
-
   // Write query result to a parquet file
   result_table->to_parquet("q6.parquet");
   return 0;

diff --git a/cpp/examples/tpch/q9.cpp → cpp/benchmarks/tpch/q9.cpp b/cpp/examples/tpch/q9.cpp → cpp/benchmarks/tpch/q9.cpp
@@ -14,7 +14,6 @@
  * limitations under the License.
  */
 
-#include "../utilities/timer.hpp"
 #include "utils.hpp"
 
 #include <cudf/column/column.hpp>
@@ -116,8 +115,6 @@ int main(int argc, char const** argv)
   auto resource = create_memory_resource(args.memory_resource_type);
   rmm::mr::set_current_device_resource(resource.get());
 
-  cudf::examples::timer timer;
-
   // Read out the table from parquet files
   auto const lineitem = read_parquet(
     args.dataset_dir + "/lineitem.parquet",
@@ -174,8 +171,6 @@ int main(int argc, char const** argv)
   auto const orderedby_table = apply_orderby(
     groupedby_table, {"nation", "o_year"}, {cudf::order::ASCENDING, cudf::order::DESCENDING});
 
-  timer.print_elapsed_millis();
-
   // Write query result to a parquet file
   orderedby_table->to_parquet("q9.parquet");
   return 0;

diff --git a/cpp/examples/tpch/utils.hpp → cpp/benchmarks/tpch/utils.hpp b/cpp/examples/tpch/utils.hpp → cpp/benchmarks/tpch/utils.hpp
@@ -312,7 +312,7 @@ struct groupby_context_t {
   }
   auto agg_results = groupby_obj.aggregate(requests);
   std::vector<std::unique_ptr<cudf::column>> result_columns;
-  for (size_t i = 0; i < agg_results.first->num_columns(); i++) {
+  for (auto i = 0; i < agg_results.first->num_columns(); i++) {
     auto col = std::make_unique<cudf::column>(agg_results.first->get_column(i));
     result_columns.push_back(std::move(col));
   }

diff --git a/cpp/examples/build.sh b/cpp/examples/build.sh
@@ -57,7 +57,6 @@ build_example() {
 }
 
 build_example basic
-build_example tpch
 build_example strings
 build_example nested_types
 build_example parquet_io

diff --git a/cpp/examples/tpch/CMakeLists.txt b/cpp/examples/tpch/CMakeLists.txt