Implement TPCH Query 2 in TpchQueryBuilder (facebookincubator#9825)

Summary: This PR introduces TPC-H Query 2 into the TpchQueryBuilder and extends the TpchBenchmark and ParquetTpchTest to include this query. Additionally, it provides a detailed performance comparison with DuckDB using the Parquet file format and includes the output of PrintPlanWithStats for detailed analysis. Scaling Factor used is 1. Here is the link to the PowerPoint presentation, which contains a detailed description for each driver and thread : https://ibm.box.com/s/sau464qdfac45aainwpj6pyvkvbtlsat ### Performance Comparison Chip: Apple M1 Pro Total Number of Cores: 10 (8 performance and 2 efficiency) Memory: 32 GB The following table summarizes the performance comparison between Velox and DuckDB (with Parquet file format) across various numbers of threads/drivers: | # Num Threads/ Drivers | Velox(ms) | DuckDB(ms) | |:----------------------:|:---------:|:----------:| | 1 | 27 | 88.4 | | 4 | 23 | 84.1 | | 8 | 25 | 82.8 | | 16 | 30 | 84 | Pull Request resolved: facebookincubator#9825 Reviewed By: bikramSingh91 Differential Revision: D58244304 Pulled By: kevinwilfong fbshipit-source-id: 4d216b10b49847ab692a394783bd5c59a59c9eb2
deepashreeraghu · Jun 13, 2024 · 32ad898 · 32ad898
1 parent 8b0ac85
commit 32ad898
Show file tree

Hide file tree

Showing 4 changed files with 214 additions and 0 deletions.
diff --git a/velox/benchmarks/tpch/TpchBenchmark.cpp b/velox/benchmarks/tpch/TpchBenchmark.cpp
@@ -494,6 +494,11 @@ BENCHMARK(q1) {
   benchmark.run(planContext);
 }
 
+BENCHMARK(q2) {
+  const auto planContext = queryBuilder->getQueryPlan(2);
+  benchmark.run(planContext);
+}
+
 BENCHMARK(q3) {
   const auto planContext = queryBuilder->getQueryPlan(3);
   benchmark.run(planContext);

diff --git a/velox/dwio/parquet/tests/ParquetTpchTest.cpp b/velox/dwio/parquet/tests/ParquetTpchTest.cpp
@@ -161,6 +161,11 @@ TEST_F(ParquetTpchTest, Q1) {
   assertQuery(1);
 }
 
+TEST_F(ParquetTpchTest, Q2) {
+  std::vector<uint32_t> sortingKeys{0, 1, 2, 3};
+  assertQuery(2, std::move(sortingKeys));
+}
+
 TEST_F(ParquetTpchTest, Q3) {
   std::vector<uint32_t> sortingKeys{1, 2};
   assertQuery(3, std::move(sortingKeys));

diff --git a/velox/exec/tests/utils/TpchQueryBuilder.cpp b/velox/exec/tests/utils/TpchQueryBuilder.cpp
@@ -142,6 +142,8 @@ TpchPlan TpchQueryBuilder::getQueryPlan(int queryId) const {
   switch (queryId) {
     case 1:
       return getQ1Plan();
+    case 2:
+      return getQ2Plan();
     case 3:
       return getQ3Plan();
     case 5:
@@ -238,6 +240,207 @@ TpchPlan TpchQueryBuilder::getQ1Plan() const {
   return context;
 }
 
+TpchPlan TpchQueryBuilder::getQ2Plan() const {
+  std::vector<std::string> supplierColumnsSubQuery = {
+      "s_suppkey", "s_nationkey"};
+  std::vector<std::string> nationColumnsSubQuery = {
+      "n_nationkey", "n_regionkey"};
+  std::vector<std::string> supplierColumns = {
+      "s_acctbal",
+      "s_name",
+      "s_address",
+      "s_phone",
+      "s_comment",
+      "s_suppkey",
+      "s_nationkey"};
+  std::vector<std::string> partColumns = {
+      "p_partkey", "p_mfgr", "p_size", "p_type"};
+  std::vector<std::string> partsuppColumns = {
+      "ps_partkey", "ps_suppkey", "ps_supplycost"};
+  std::vector<std::string> nationColumns = {
+      "n_nationkey", "n_name", "n_regionkey"};
+  std::vector<std::string> regionColumns = {"r_regionkey", "r_name"};
+
+  auto supplierSelectedRowTypeSubQuery =
+      getRowType(kSupplier, supplierColumnsSubQuery);
+  const auto& supplierFileColumnsSubQuery = getFileColumnNames(kSupplier);
+  auto nationSelectedRowTypeSubQuery =
+      getRowType(kNation, nationColumnsSubQuery);
+  const auto& nationFileColumnsSubQuery = getFileColumnNames(kNation);
+  auto partSelectedRowType = getRowType(kPart, partColumns);
+  const auto& partFileColumns = getFileColumnNames(kPart);
+  auto supplierSelectedRowType = getRowType(kSupplier, supplierColumns);
+  const auto& supplierFileColumns = getFileColumnNames(kSupplier);
+  auto partsuppSelectedRowType = getRowType(kPartsupp, partsuppColumns);
+  const auto& partsuppFileColumns = getFileColumnNames(kPartsupp);
+  auto nationSelectedRowType = getRowType(kNation, nationColumns);
+  const auto& nationFileColumns = getFileColumnNames(kNation);
+  auto regionSelectedRowType = getRowType(kRegion, regionColumns);
+  const auto& regionFileColumns = getFileColumnNames(kRegion);
+
+  const std::string regionNameFilter = "r_name = 'EUROPE'";
+
+  auto planNodeIdGenerator = std::make_shared<core::PlanNodeIdGenerator>();
+  core::PlanNodeId supplierScanIdSubQuery;
+  core::PlanNodeId partsuppScanIdSubQuery;
+  core::PlanNodeId nationScanIdSubQuery;
+  core::PlanNodeId regionScanIdSubQuery;
+  core::PlanNodeId partScanId;
+  core::PlanNodeId supplierScanId;
+  core::PlanNodeId partsuppScanId;
+  core::PlanNodeId nationScanId;
+  core::PlanNodeId regionScanId;
+
+  auto regionSubQuery = PlanBuilder(planNodeIdGenerator)
+                            .tableScan(
+                                kRegion,
+                                regionSelectedRowType,
+                                regionFileColumns,
+                                {regionNameFilter})
+                            .capturePlanNodeId(regionScanIdSubQuery)
+                            .planNode();
+
+  auto nationJoinRegionSubQuery =
+      PlanBuilder(planNodeIdGenerator)
+          .tableScan(
+              kNation, nationSelectedRowTypeSubQuery, nationFileColumnsSubQuery)
+          .capturePlanNodeId(nationScanIdSubQuery)
+          .hashJoin(
+              {"n_regionkey"},
+              {"r_regionkey"},
+              regionSubQuery,
+              "",
+              {"n_nationkey"})
+          .planNode();
+
+  auto supplierJoinNationJoinRegionSubQuery =
+      PlanBuilder(planNodeIdGenerator)
+          .tableScan(
+              kSupplier,
+              supplierSelectedRowTypeSubQuery,
+              supplierFileColumnsSubQuery)
+          .capturePlanNodeId(supplierScanIdSubQuery)
+          .hashJoin(
+              {"s_nationkey"},
+              {"n_nationkey"},
+              nationJoinRegionSubQuery,
+              "",
+              {"s_suppkey"})
+          .planNode();
+
+  auto part = PlanBuilder(planNodeIdGenerator)
+                  .tableScan(
+                      kPart,
+                      partSelectedRowType,
+                      partFileColumns,
+                      {},
+                      "p_type like '%BRASS'")
+                  .capturePlanNodeId(partScanId)
+                  .filter("p_size = 15")
+                  .planNode();
+
+  auto region = PlanBuilder(planNodeIdGenerator)
+                    .tableScan(
+                        kRegion,
+                        regionSelectedRowType,
+                        regionFileColumns,
+                        {regionNameFilter})
+                    .capturePlanNodeId(regionScanId)
+                    .planNode();
+
+  auto nationJoinRegion =
+      PlanBuilder(planNodeIdGenerator)
+          .tableScan(kNation, nationSelectedRowType, nationFileColumns)
+          .capturePlanNodeId(nationScanId)
+          .hashJoin(
+              {"n_regionkey"},
+              {"r_regionkey"},
+              region,
+              "",
+              {"n_nationkey", "n_name"})
+          .planNode();
+
+  auto supplierJoinNationJoinRegion =
+      PlanBuilder(planNodeIdGenerator)
+          .tableScan(kSupplier, supplierSelectedRowType, supplierFileColumns)
+          .capturePlanNodeId(supplierScanId)
+          .hashJoin(
+              {"s_nationkey"},
+              {"n_nationkey"},
+              nationJoinRegion,
+              "",
+              mergeColumnNames(supplierColumns, {"s_suppkey", "n_name"}))
+          .planNode();
+
+  auto partsuppJoinPartJoinSupplierJoinNationJoinRegion =
+      PlanBuilder(planNodeIdGenerator)
+          .tableScan(kPartsupp, partsuppSelectedRowType, partsuppFileColumns)
+          .capturePlanNodeId(partsuppScanId)
+          .hashJoin(
+              {"ps_partkey"},
+              {"p_partkey"},
+              part,
+              "",
+              {"ps_suppkey", "ps_supplycost", "p_partkey", "p_mfgr"})
+          .hashJoin(
+              {"ps_suppkey"},
+              {"s_suppkey"},
+              supplierJoinNationJoinRegion,
+              "",
+              mergeColumnNames(
+                  supplierColumns,
+                  {"ps_supplycost", "p_partkey", "p_mfgr", "n_name"}))
+          .planNode();
+
+  auto plan =
+      PlanBuilder(planNodeIdGenerator)
+          .tableScan(kPartsupp, partsuppSelectedRowType, partsuppFileColumns)
+          .capturePlanNodeId(partsuppScanIdSubQuery)
+          .hashJoin(
+              {"ps_suppkey"},
+              {"s_suppkey"},
+              supplierJoinNationJoinRegionSubQuery,
+              "",
+              {"ps_supplycost", "ps_partkey"})
+          .partialAggregation(
+              {"ps_partkey"}, {"min(ps_supplycost) AS min_supplycost"})
+          .localPartition({"ps_partkey"})
+          .finalAggregation()
+          .hashJoin(
+              {"ps_partkey"},
+              {"p_partkey"},
+              partsuppJoinPartJoinSupplierJoinNationJoinRegion,
+              "ps_supplycost = min_supplycost",
+              mergeColumnNames(
+                  supplierColumns, {"p_partkey", "p_mfgr", "n_name"}))
+          .orderBy({"s_acctbal DESC", "n_name", "s_name", "p_partkey"}, false)
+          .project(
+              {"s_acctbal",
+               "s_name",
+               "n_name",
+               "p_partkey",
+               "p_mfgr",
+               "s_address",
+               "s_phone",
+               "s_comment"})
+          .limit(0, 100, false)
+          .planNode();
+
+  TpchPlan context;
+  context.plan = std::move(plan);
+  context.dataFiles[supplierScanIdSubQuery] = getTableFilePaths(kSupplier);
+  context.dataFiles[partsuppScanIdSubQuery] = getTableFilePaths(kPartsupp);
+  context.dataFiles[nationScanIdSubQuery] = getTableFilePaths(kNation);
+  context.dataFiles[regionScanIdSubQuery] = getTableFilePaths(kRegion);
+  context.dataFiles[partScanId] = getTableFilePaths(kPart);
+  context.dataFiles[supplierScanId] = getTableFilePaths(kSupplier);
+  context.dataFiles[partsuppScanId] = getTableFilePaths(kPartsupp);
+  context.dataFiles[nationScanId] = getTableFilePaths(kNation);
+  context.dataFiles[regionScanId] = getTableFilePaths(kRegion);
+  context.dataFileFormat = format_;
+  return context;
+}
+
 TpchPlan TpchQueryBuilder::getQ3Plan() const {
   std::vector<std::string> lineitemColumns = {
       "l_shipdate", "l_orderkey", "l_extendedprice", "l_discount"};

diff --git a/velox/exec/tests/utils/TpchQueryBuilder.h b/velox/exec/tests/utils/TpchQueryBuilder.h
@@ -91,6 +91,7 @@ class TpchQueryBuilder {
       const std::vector<std::string>& columns);
 
   TpchPlan getQ1Plan() const;
+  TpchPlan getQ2Plan() const;
   TpchPlan getQ3Plan() const;
   TpchPlan getQ5Plan() const;
   TpchPlan getQ6Plan() const;