| 1 | #include "benchmark_runner.hpp" |
| 2 | #include "duckdb_benchmark_macro.hpp" |
| 3 | #include "duckdb/main/appender.hpp" |
| 4 | |
| 5 | #include <random> |
| 6 | |
| 7 | using namespace duckdb; |
| 8 | using namespace std; |
| 9 | |
| 10 | #define IN_LIST_ROW_COUNT 1000000 |
| 11 | #define STRING_LENGTH 4 |
| 12 | |
| 13 | #define IN_QUERY_BODY(INCOUNT, NOT_IN) \ |
| 14 | static constexpr const char *chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"; \ |
| 15 | string in_list; \ |
| 16 | static string GenerateString(std::uniform_int_distribution<> &distribution, std::mt19937 &gen) { \ |
| 17 | string result; \ |
| 18 | for (size_t i = 0; i < STRING_LENGTH; i++) { \ |
| 19 | result += string(1, chars[distribution(gen)]); \ |
| 20 | } \ |
| 21 | return result; \ |
| 22 | } \ |
| 23 | virtual void Load(DuckDBBenchmarkState *state) { \ |
| 24 | std::uniform_int_distribution<> distribution(0, strlen(chars) - 1); \ |
| 25 | std::mt19937 gen; \ |
| 26 | gen.seed(42); \ |
| 27 | state->conn.Query("CREATE TABLE strings(s VARCHAR);"); \ |
| 28 | Appender appender(state->conn, "strings"); \ |
| 29 | for (size_t i = 0; i < IN_LIST_ROW_COUNT; i++) { \ |
| 30 | appender.BeginRow(); \ |
| 31 | appender.Append<Value>(Value(GenerateString(distribution, gen))); \ |
| 32 | appender.EndRow(); \ |
| 33 | } \ |
| 34 | appender.Close(); \ |
| 35 | for (size_t i = 0; i < INCOUNT; i++) { \ |
| 36 | in_list += "'" + GenerateString(distribution, gen) + "'"; \ |
| 37 | if (i != INCOUNT - 1) { \ |
| 38 | in_list += ", "; \ |
| 39 | } \ |
| 40 | } \ |
| 41 | } \ |
| 42 | virtual string GetQuery() { \ |
| 43 | return "SELECT * FROM strings WHERE s " + (NOT_IN ? string("NOT ") : string("")) + "IN (" + in_list + ")"; \ |
| 44 | } \ |
| 45 | virtual string VerifyResult(QueryResult *result) { \ |
| 46 | if (!result->success) { \ |
| 47 | return result->error; \ |
| 48 | } \ |
| 49 | return string(); \ |
| 50 | } \ |
| 51 | virtual string BenchmarkInfo() { \ |
| 52 | return StringUtil::Format("Runs the following query: \"" + GetQuery() + "\""); \ |
| 53 | } |
| 54 | |
| 55 | DUCKDB_BENCHMARK(InList0001Entry, "[in]" ) |
| 56 | IN_QUERY_BODY(1, false) |
| 57 | FINISH_BENCHMARK(InList0001Entry) |
| 58 | |
| 59 | DUCKDB_BENCHMARK(InList0002Entry, "[in]" ) |
| 60 | IN_QUERY_BODY(2, false) |
| 61 | FINISH_BENCHMARK(InList0002Entry) |
| 62 | |
| 63 | DUCKDB_BENCHMARK(InList0004Entry, "[in]" ) |
| 64 | IN_QUERY_BODY(4, false) |
| 65 | FINISH_BENCHMARK(InList0004Entry) |
| 66 | |
| 67 | DUCKDB_BENCHMARK(InList0008Entry, "[in]" ) |
| 68 | IN_QUERY_BODY(8, false) |
| 69 | FINISH_BENCHMARK(InList0008Entry) |
| 70 | |
| 71 | DUCKDB_BENCHMARK(InList0016Entry, "[in]" ) |
| 72 | IN_QUERY_BODY(16, false) |
| 73 | FINISH_BENCHMARK(InList0016Entry) |
| 74 | |
| 75 | DUCKDB_BENCHMARK(InList0032Entry, "[in]" ) |
| 76 | IN_QUERY_BODY(32, false) |
| 77 | FINISH_BENCHMARK(InList0032Entry) |
| 78 | |
| 79 | DUCKDB_BENCHMARK(InList0064Entry, "[in]" ) |
| 80 | IN_QUERY_BODY(64, false) |
| 81 | FINISH_BENCHMARK(InList0064Entry) |
| 82 | |
| 83 | DUCKDB_BENCHMARK(InList0128Entry, "[in]" ) |
| 84 | IN_QUERY_BODY(128, false) |
| 85 | FINISH_BENCHMARK(InList0128Entry) |
| 86 | |
| 87 | DUCKDB_BENCHMARK(InList0256Entry, "[in]" ) |
| 88 | IN_QUERY_BODY(256, false) |
| 89 | FINISH_BENCHMARK(InList0256Entry) |
| 90 | |
| 91 | DUCKDB_BENCHMARK(InList0512Entry, "[in]" ) |
| 92 | IN_QUERY_BODY(512, false) |
| 93 | FINISH_BENCHMARK(InList0512Entry) |
| 94 | |
| 95 | DUCKDB_BENCHMARK(InList1024Entry, "[in]" ) |
| 96 | IN_QUERY_BODY(1024, false) |
| 97 | FINISH_BENCHMARK(InList1024Entry) |
| 98 | |
| 99 | DUCKDB_BENCHMARK(InList2048Entry, "[in]" ) |
| 100 | IN_QUERY_BODY(2048, false) |
| 101 | FINISH_BENCHMARK(InList2048Entry) |
| 102 | |
| 103 | DUCKDB_BENCHMARK(NotInList0001Entry, "[in]" ) |
| 104 | IN_QUERY_BODY(1, true) |
| 105 | FINISH_BENCHMARK(NotInList0001Entry) |
| 106 | |
| 107 | DUCKDB_BENCHMARK(NotInList0002Entry, "[in]" ) |
| 108 | IN_QUERY_BODY(2, true) |
| 109 | FINISH_BENCHMARK(NotInList0002Entry) |
| 110 | |
| 111 | DUCKDB_BENCHMARK(NotInList0004Entry, "[in]" ) |
| 112 | IN_QUERY_BODY(4, true) |
| 113 | FINISH_BENCHMARK(NotInList0004Entry) |
| 114 | |
| 115 | DUCKDB_BENCHMARK(NotInList0008Entry, "[in]" ) |
| 116 | IN_QUERY_BODY(8, true) |
| 117 | FINISH_BENCHMARK(NotInList0008Entry) |
| 118 | |
| 119 | DUCKDB_BENCHMARK(NotInList0016Entry, "[in]" ) |
| 120 | IN_QUERY_BODY(16, true) |
| 121 | FINISH_BENCHMARK(NotInList0016Entry) |
| 122 | |
| 123 | DUCKDB_BENCHMARK(NotInList0032Entry, "[in]" ) |
| 124 | IN_QUERY_BODY(32, true) |
| 125 | FINISH_BENCHMARK(NotInList0032Entry) |
| 126 | |
| 127 | DUCKDB_BENCHMARK(NotInList0064Entry, "[in]" ) |
| 128 | IN_QUERY_BODY(64, true) |
| 129 | FINISH_BENCHMARK(NotInList0064Entry) |
| 130 | |
| 131 | DUCKDB_BENCHMARK(NotInList0128Entry, "[in]" ) |
| 132 | IN_QUERY_BODY(128, true) |
| 133 | FINISH_BENCHMARK(NotInList0128Entry) |
| 134 | |
| 135 | DUCKDB_BENCHMARK(NotInList0256Entry, "[in]" ) |
| 136 | IN_QUERY_BODY(256, true) |
| 137 | FINISH_BENCHMARK(NotInList0256Entry) |
| 138 | |
| 139 | DUCKDB_BENCHMARK(NotInList0512Entry, "[in]" ) |
| 140 | IN_QUERY_BODY(512, true) |
| 141 | FINISH_BENCHMARK(NotInList0512Entry) |
| 142 | |
| 143 | DUCKDB_BENCHMARK(NotInList1024Entry, "[in]" ) |
| 144 | IN_QUERY_BODY(1024, true) |
| 145 | FINISH_BENCHMARK(NotInList1024Entry) |
| 146 | |
| 147 | DUCKDB_BENCHMARK(NotInList2048Entry, "[in]" ) |
| 148 | IN_QUERY_BODY(2048, true) |
| 149 | FINISH_BENCHMARK(NotInList2048Entry) |
| 150 | |