Add scale_factor axes to queries

cpp/benchmarks/tpch/q1.cpp

@@ -103,9 +103,11 @@
 
 void run_tpch_q1(nvbench::state& state)
 {
+  double const scale_factor = state.get_float64("scale_factor");
+
   // Define a map for holding parquet data sources
   std::unordered_map<std::string, parquet_device_buffer> sources;
-  generate_parquet_data_sources({"lineitem"}, sources);
+  generate_parquet_data_sources(scale_factor, {"lineitem"}, sources);
 
   // Define the column projections and filter predicate for `lineitem` table
   std::vector<std::string> const lineitem_cols = {"l_returnflag",
@@ -174,4 +176,4 @@ void tpch_q1(nvbench::state& state)
   state.exec(nvbench::exec_tag::sync, [&](nvbench::launch& launch) { run_tpch_q1(state); });
 }
 
-NVBENCH_BENCH(tpch_q1).set_name("tpch_q1");
+NVBENCH_BENCH(tpch_q1).set_name("tpch_q1").add_float64_axis("scale_factor", {0.01, 0.1, 1});

cpp/benchmarks/tpch/q10.cpp

@@ -93,9 +93,12 @@
 
 void run_tpch_q10(nvbench::state& state)
 {
+  double const scale_factor = state.get_float64("scale_factor");
+
   // Define a map for holding parquet data sources
   std::unordered_map<std::string, parquet_device_buffer> sources;
-  generate_parquet_data_sources({"customer", "orders", "lineitem", "nation"}, sources);
+  generate_parquet_data_sources(
+    scale_factor, {"customer", "orders", "lineitem", "nation"}, sources);
 
   // Define the column projection and filter predicate for the `orders` table
   std::vector<std::string> const orders_cols = {"o_custkey", "o_orderkey", "o_orderdate"};
@@ -167,4 +170,4 @@ void tpch_q10(nvbench::state& state)
   state.exec(nvbench::exec_tag::sync, [&](nvbench::launch& launch) { run_tpch_q10(state); });
 }
 
-NVBENCH_BENCH(tpch_q10).set_name("tpch_q10");
+NVBENCH_BENCH(tpch_q10).set_name("tpch_q10").add_float64_axis("scale_factor", {0.01, 0.1, 1});

cpp/benchmarks/tpch/q5.cpp

@@ -88,10 +88,12 @@
 
 void run_tpch_q5(nvbench::state& state)
 {
+  double const scale_factor = state.get_float64("scale_factor");
+
   // Define a map for holding parquet data sources
   std::unordered_map<std::string, parquet_device_buffer> sources;
-  generate_parquet_data_sources({"customer", "orders", "lineitem", "supplier", "nation", "region"},
-                                sources);
+  generate_parquet_data_sources(
+    scale_factor, {"customer", "orders", "lineitem", "supplier", "nation", "region"}, sources);
 
   // Define the column projection and filter predicate for the `orders` table
   std::vector<std::string> const orders_cols = {"o_custkey", "o_orderkey", "o_orderdate"};
@@ -170,4 +172,4 @@ void tpch_q5(nvbench::state& state)
   state.exec(nvbench::exec_tag::sync, [&](nvbench::launch& launch) { run_tpch_q5(state); });
 }
 
-NVBENCH_BENCH(tpch_q5).set_name("tpch_q5");
+NVBENCH_BENCH(tpch_q5).set_name("tpch_q5").add_float64_axis("scale_factor", {0.01, 0.1, 1});

cpp/benchmarks/tpch/q6.cpp

@@ -62,9 +62,11 @@
 
 void run_tpch_q6(nvbench::state& state)
 {
+  double const scale_factor = state.get_float64("scale_factor");
+
   // Define a map for holding parquet data sources
   std::unordered_map<std::string, parquet_device_buffer> sources;
-  generate_parquet_data_sources({"lineitem"}, sources);
+  generate_parquet_data_sources(scale_factor, {"lineitem"}, sources);
 
   // Read out the `lineitem` table from parquet file
   std::vector<std::string> const lineitem_cols = {
@@ -137,4 +139,4 @@ void tpch_q6(nvbench::state& state)
   state.exec(nvbench::exec_tag::sync, [&](nvbench::launch& launch) { run_tpch_q6(state); });
 }
 
-NVBENCH_BENCH(tpch_q6).set_name("tpch_q6");
+NVBENCH_BENCH(tpch_q6).set_name("tpch_q6").add_float64_axis("scale_factor", {0.01, 0.1, 1});

cpp/benchmarks/tpch/q9.cpp

@@ -111,10 +111,12 @@
 
 void run_tpch_q9(nvbench::state& state)
 {
+  double const scale_factor = state.get_float64("scale_factor");
+
   // Define a map for holding parquet data sources
   std::unordered_map<std::string, parquet_device_buffer> sources;
-  generate_parquet_data_sources({"part", "supplier", "lineitem", "partsupp", "orders", "nation"},
-                                sources);
+  generate_parquet_data_sources(
+    scale_factor, {"part", "supplier", "lineitem", "partsupp", "orders", "nation"}, sources);
 
   // Read out the table from parquet files
   auto const lineitem = read_parquet(
@@ -183,4 +185,4 @@ void tpch_q9(nvbench::state& state)
   state.exec(nvbench::exec_tag::sync, [&](nvbench::launch& launch) { run_tpch_q9(state); });
 }
 
-NVBENCH_BENCH(tpch_q9).set_name("tpch_q9");
+NVBENCH_BENCH(tpch_q9).set_name("tpch_q9").add_float64_axis("scale_factor", {0.01, 0.1, 1});

cpp/benchmarks/tpch/utils.hpp

@@ -526,17 +526,6 @@ int32_t days_since_epoch(int year, int month, int day)
   return static_cast<int32_t>(diff);
 }
 
-/**
- * @brief Read the scale factor from the environment variable `CUDF_TPCH_SF`
- */
-double get_sf()
-{
-  char* val       = getenv("CUDF_TPCH_SF");
-  double const sf = (val == nullptr) ? 1 : std::atof(val);
-  std::cout << "Using scale factor: " << sf << std::endl;
-  return sf;
-}
-
 /**
  * @brief Struct representing a parquet device buffer
  */
@@ -588,19 +577,19 @@ void write_to_parquet_device_buffer(std::unique_ptr<cudf::table> const& table,
 /**
  * @brief Generate TPC-H tables and write to parquet device buffers
  *
+ * @param scale_factor The scale factor of NDS-H tables to generate
  * @param table_names The names of the tables to generate
  * @param sources The parquet data sources to populate
  */
-void generate_parquet_data_sources(std::vector<std::string> const& table_names,
+void generate_parquet_data_sources(double scale_factor,
+                                   std::vector<std::string> const& table_names,
                                    std::unordered_map<std::string, parquet_device_buffer>& sources)
 {
   CUDF_FUNC_RANGE();
   std::for_each(table_names.begin(), table_names.end(), [&](auto const& table_name) {
     sources[table_name] = parquet_device_buffer();
   });
 
-  auto scale_factor = get_sf();
-
   auto [orders, lineitem, part] = cudf::datagen::generate_orders_lineitem_part(
     scale_factor, cudf::get_default_stream(), rmm::mr::get_current_device_resource());