| 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 STRING_COUNT 10000000 |
| 11 | #define STRING_LENGTH 4 |
| 12 | |
| 13 | #define STRING_DATA_GEN_BODY(STRING_LENGTH) \ |
| 14 | static constexpr const char *chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"; \ |
| 15 | static string GenerateString(std::uniform_int_distribution<> &distribution, std::mt19937 &gen) { \ |
| 16 | string result; \ |
| 17 | for (size_t i = 0; i < STRING_LENGTH; i++) { \ |
| 18 | result += string(1, chars[distribution(gen)]); \ |
| 19 | } \ |
| 20 | return result; \ |
| 21 | } \ |
| 22 | void Load(DuckDBBenchmarkState *state) override { \ |
| 23 | std::uniform_int_distribution<> distribution(0, strlen(chars) - 1); \ |
| 24 | std::mt19937 gen; \ |
| 25 | gen.seed(42); \ |
| 26 | state->conn.Query("CREATE TABLE strings(s1 VARCHAR, s2 VARCHAR);"); \ |
| 27 | Appender appender(state->conn, "strings"); \ |
| 28 | for (size_t i = 0; i < STRING_COUNT; i++) { \ |
| 29 | appender.BeginRow(); \ |
| 30 | appender.Append<Value>(Value(GenerateString(distribution, gen))); \ |
| 31 | appender.Append<Value>(Value(GenerateString(distribution, gen))); \ |
| 32 | appender.EndRow(); \ |
| 33 | } \ |
| 34 | appender.Close(); \ |
| 35 | } \ |
| 36 | string VerifyResult(QueryResult *result) override { \ |
| 37 | if (!result->success) { \ |
| 38 | return result->error; \ |
| 39 | } \ |
| 40 | return string(); \ |
| 41 | } |
| 42 | |
| 43 | DUCKDB_BENCHMARK(StringConcatShort, "[string]" ) |
| 44 | STRING_DATA_GEN_BODY(4) |
| 45 | string GetQuery() override { |
| 46 | return "SELECT s1 || s2 FROM strings" ; |
| 47 | } |
| 48 | string BenchmarkInfo() override { |
| 49 | return "STRING CONCAT" ; |
| 50 | } |
| 51 | FINISH_BENCHMARK(StringConcatShort) |
| 52 | |
| 53 | DUCKDB_BENCHMARK(StringConcatLong, "[string]" ) |
| 54 | STRING_DATA_GEN_BODY(20) |
| 55 | string GetQuery() override { |
| 56 | return "SELECT s1 || s2 FROM strings" ; |
| 57 | } |
| 58 | string BenchmarkInfo() override { |
| 59 | return "STRING CONCAT" ; |
| 60 | } |
| 61 | FINISH_BENCHMARK(StringConcatLong) |
| 62 | |
| 63 | DUCKDB_BENCHMARK(StringEqualityShort, "[string]" ) |
| 64 | STRING_DATA_GEN_BODY(4) |
| 65 | string GetQuery() override { |
| 66 | return "SELECT s1=s2 FROM strings" ; |
| 67 | } |
| 68 | string BenchmarkInfo() override { |
| 69 | return "STRING COMPARISON" ; |
| 70 | } |
| 71 | FINISH_BENCHMARK(StringEqualityShort) |
| 72 | |
| 73 | DUCKDB_BENCHMARK(StringEqualityLong, "[string]" ) |
| 74 | STRING_DATA_GEN_BODY(20) |
| 75 | string GetQuery() override { |
| 76 | return "SELECT s1=s2 FROM strings" ; |
| 77 | } |
| 78 | string BenchmarkInfo() override { |
| 79 | return "STRING COMPARISON" ; |
| 80 | } |
| 81 | FINISH_BENCHMARK(StringEqualityLong) |
| 82 | |
| 83 | DUCKDB_BENCHMARK(StringGreaterThanShort, "[string]" ) |
| 84 | STRING_DATA_GEN_BODY(4) |
| 85 | string GetQuery() override { |
| 86 | return "SELECT s1>s2 FROM strings" ; |
| 87 | } |
| 88 | string BenchmarkInfo() override { |
| 89 | return "STRING COMPARISON" ; |
| 90 | } |
| 91 | FINISH_BENCHMARK(StringGreaterThanShort) |
| 92 | |
| 93 | DUCKDB_BENCHMARK(StringGreaterThanLong, "[string]" ) |
| 94 | STRING_DATA_GEN_BODY(20) |
| 95 | string GetQuery() override { |
| 96 | return "SELECT s1>s2 FROM strings" ; |
| 97 | } |
| 98 | string BenchmarkInfo() override { |
| 99 | return "STRING COMPARISON" ; |
| 100 | } |
| 101 | FINISH_BENCHMARK(StringGreaterThanLong) |
| 102 | |
| 103 | DUCKDB_BENCHMARK(StringLengthShort, "[string]" ) |
| 104 | STRING_DATA_GEN_BODY(5) |
| 105 | string GetQuery() override { |
| 106 | return "SELECT LENGTH(s1)+LENGTH(s2) FROM strings" ; |
| 107 | } |
| 108 | string BenchmarkInfo() override { |
| 109 | return "STRING LENGTH" ; |
| 110 | } |
| 111 | FINISH_BENCHMARK(StringLengthShort) |
| 112 | |
| 113 | DUCKDB_BENCHMARK(StringLengthLong, "[string]" ) |
| 114 | STRING_DATA_GEN_BODY(20) |
| 115 | string GetQuery() override { |
| 116 | return "SELECT LENGTH(s1)+LENGTH(s2) FROM strings" ; |
| 117 | } |
| 118 | string BenchmarkInfo() override { |
| 119 | return "STRING LENGTH" ; |
| 120 | } |
| 121 | FINISH_BENCHMARK(StringLengthLong) |
| 122 | |
| 123 | DUCKDB_BENCHMARK(StringAggShort, "[string]" ) |
| 124 | STRING_DATA_GEN_BODY(4) |
| 125 | string GetQuery() override { |
| 126 | return "SELECT STRING_AGG(s1, ' ') FROM strings" ; |
| 127 | } |
| 128 | string BenchmarkInfo() override { |
| 129 | return "STRING LENGTH" ; |
| 130 | } |
| 131 | FINISH_BENCHMARK(StringAggShort) |
| 132 | |
| 133 | DUCKDB_BENCHMARK(StringAggLong, "[string]" ) |
| 134 | STRING_DATA_GEN_BODY(20) |
| 135 | string GetQuery() override { |
| 136 | return "SELECT STRING_AGG(s1, ' ') FROM strings" ; |
| 137 | } |
| 138 | string BenchmarkInfo() override { |
| 139 | return "STRING LENGTH" ; |
| 140 | } |
| 141 | FINISH_BENCHMARK(StringAggLong) |
| 142 | |
| 143 | DUCKDB_BENCHMARK(StringInstr, "[string]" ) |
| 144 | STRING_DATA_GEN_BODY(4) |
| 145 | string GetQuery() override { |
| 146 | return "SELECT INSTR(s1, 'h') FROM strings" ; |
| 147 | } |
| 148 | string BenchmarkInfo() override { |
| 149 | return "STRING INSTR" ; |
| 150 | } |
| 151 | FINISH_BENCHMARK(StringInstr) |
| 152 | |
| 153 | DUCKDB_BENCHMARK(StringInstrNull, "[string]" ) |
| 154 | STRING_DATA_GEN_BODY(4) |
| 155 | string GetQuery() override { |
| 156 | return "SELECT INSTR(s1, '') FROM strings" ; |
| 157 | } |
| 158 | string BenchmarkInfo() override { |
| 159 | return "STRING INSTR" ; |
| 160 | } |
| 161 | FINISH_BENCHMARK(StringInstrNull) |
| 162 | |
| 163 | //------------------------- CONTAINS ----------------------------------------- |
| 164 | DUCKDB_BENCHMARK(StringContains, "[string]" ) |
| 165 | STRING_DATA_GEN_BODY(4) |
| 166 | string GetQuery() override { |
| 167 | return "SELECT CONTAINS(s1, 'h') FROM strings" ; |
| 168 | } |
| 169 | string BenchmarkInfo() override { |
| 170 | return "STRING CONTAINS" ; |
| 171 | } |
| 172 | FINISH_BENCHMARK(StringContains) |
| 173 | |
| 174 | DUCKDB_BENCHMARK(StringContainsNull, "[string]" ) |
| 175 | STRING_DATA_GEN_BODY(4) |
| 176 | string GetQuery() override { |
| 177 | return "SELECT CONTAINS(s1, '') FROM strings" ; |
| 178 | } |
| 179 | string BenchmarkInfo() override { |
| 180 | return "STRING CONTAINS" ; |
| 181 | } |
| 182 | FINISH_BENCHMARK(StringContainsNull) |
| 183 | //----------------------------------------------------------------------------- |
| 184 | |
| 185 | //------------------------- CONTAINS LIKE ------------------------------------- |
| 186 | DUCKDB_BENCHMARK(StringContainsLike, "[string]" ) |
| 187 | STRING_DATA_GEN_BODY(4) |
| 188 | string GetQuery() override { |
| 189 | return "SELECT s1 FROM strings WHERE s1 LIKE '%h%'" ; |
| 190 | } |
| 191 | string BenchmarkInfo() override { |
| 192 | return "STRING CONTAINS LIKE" ; |
| 193 | } |
| 194 | FINISH_BENCHMARK(StringContainsLike) |
| 195 | //----------------------------------------------------------------------------- |
| 196 | |
| 197 | DUCKDB_BENCHMARK(StringRegex, "[string]" ) |
| 198 | STRING_DATA_GEN_BODY(4) |
| 199 | string GetQuery() override { |
| 200 | return "SELECT REGEXP_MATCHES(s1, 'h') FROM strings" ; |
| 201 | } |
| 202 | string BenchmarkInfo() override { |
| 203 | return "STRING REGEX" ; |
| 204 | } |
| 205 | FINISH_BENCHMARK(StringRegex) |
| 206 | |
| 207 | DUCKDB_BENCHMARK(StringRegexNull, "[string]" ) |
| 208 | STRING_DATA_GEN_BODY(4) |
| 209 | string GetQuery() override { |
| 210 | return "SELECT REGEXP_MATCHES(s1, '') FROM strings" ; |
| 211 | } |
| 212 | string BenchmarkInfo() override { |
| 213 | return "STRING REGEX" ; |
| 214 | } |
| 215 | FINISH_BENCHMARK(StringRegexNull) |
| 216 | |
| 217 | //----------------------- PREFIX benchmark ----------------------------------// |
| 218 | DUCKDB_BENCHMARK(StringPrefix, "[string]" ) |
| 219 | STRING_DATA_GEN_BODY(4) |
| 220 | string GetQuery() override { |
| 221 | return "SELECT prefix(s1, 'a') FROM strings" ; |
| 222 | } |
| 223 | string BenchmarkInfo() override { |
| 224 | return "STRING PREFIX early out" ; |
| 225 | } |
| 226 | FINISH_BENCHMARK(StringPrefix) |
| 227 | |
| 228 | DUCKDB_BENCHMARK(StringPrefixInlined, "[string]" ) |
| 229 | STRING_DATA_GEN_BODY(12) |
| 230 | string GetQuery() override { |
| 231 | return "SELECT prefix(s1, 'a') FROM strings" ; |
| 232 | } |
| 233 | string BenchmarkInfo() override { |
| 234 | return "STRING PREFIX inlined" ; |
| 235 | } |
| 236 | FINISH_BENCHMARK(StringPrefixInlined) |
| 237 | |
| 238 | DUCKDB_BENCHMARK(StringPrefixPointer, "[string]" ) |
| 239 | STRING_DATA_GEN_BODY(20) |
| 240 | string GetQuery() override { |
| 241 | return "SELECT prefix(s1, 'a') FROM strings" ; |
| 242 | } |
| 243 | string BenchmarkInfo() override { |
| 244 | return "STRING PREFIX store pointer" ; |
| 245 | } |
| 246 | FINISH_BENCHMARK(StringPrefixPointer) |
| 247 | |
| 248 | //----------------------- PREFIX REGEX benchmark ----------------------------------// |
| 249 | DUCKDB_BENCHMARK(StringPrefixRegEX, "[string]" ) |
| 250 | STRING_DATA_GEN_BODY(4) |
| 251 | string GetQuery() override { |
| 252 | return "SELECT REGEXP_MATCHES(s1, 'a%') FROM strings" ; |
| 253 | } |
| 254 | string BenchmarkInfo() override { |
| 255 | return "STRING PREFIX REGEX" ; |
| 256 | } |
| 257 | FINISH_BENCHMARK(StringPrefixRegEX) |
| 258 | |
| 259 | //----------------------- PREFIX LIKE benchmark ----------------------------------// |
| 260 | DUCKDB_BENCHMARK(StringPrefixLike, "[string]" ) |
| 261 | STRING_DATA_GEN_BODY(4) |
| 262 | string GetQuery() override { |
| 263 | return "SELECT s1 FROM strings WHERE s1 LIKE 'a%'" ; |
| 264 | } |
| 265 | string BenchmarkInfo() override { |
| 266 | return "STRING PREFIX LIKE" ; |
| 267 | } |
| 268 | FINISH_BENCHMARK(StringPrefixLike) |
| 269 | |
| 270 | DUCKDB_BENCHMARK(StringPrefixInlinedLike, "[string]" ) |
| 271 | STRING_DATA_GEN_BODY(12) |
| 272 | string GetQuery() override { |
| 273 | return "SELECT s1 FROM strings WHERE s1 LIKE 'a%'" ; |
| 274 | } |
| 275 | string BenchmarkInfo() override { |
| 276 | return "STRING PREFIX inlined LIKE" ; |
| 277 | } |
| 278 | FINISH_BENCHMARK(StringPrefixInlinedLike) |
| 279 | |
| 280 | DUCKDB_BENCHMARK(StringPrefixPointerLike, "[string]" ) |
| 281 | STRING_DATA_GEN_BODY(20) |
| 282 | string GetQuery() override { |
| 283 | return "SELECT s1 FROM strings WHERE s1 LIKE 'a%'" ; |
| 284 | } |
| 285 | string BenchmarkInfo() override { |
| 286 | return "STRING PREFIX store pointer LIKE" ; |
| 287 | } |
| 288 | FINISH_BENCHMARK(StringPrefixPointerLike) |
| 289 | |
| 290 | //----------------------- SUFFIX benchmark ----------------------------------// |
| 291 | DUCKDB_BENCHMARK(StringSuffixShort, "[string]" ) |
| 292 | STRING_DATA_GEN_BODY(4) |
| 293 | string GetQuery() override { |
| 294 | return "SELECT suffix(s1, 'a') FROM strings" ; |
| 295 | } |
| 296 | string BenchmarkInfo() override { |
| 297 | return "Short string for suffix" ; |
| 298 | } |
| 299 | FINISH_BENCHMARK(StringSuffixShort) |
| 300 | |
| 301 | DUCKDB_BENCHMARK(StringSuffixLong, "[string]" ) |
| 302 | STRING_DATA_GEN_BODY(20) |
| 303 | string GetQuery() override { |
| 304 | return "SELECT suffix(s1, 'a') FROM strings" ; |
| 305 | } |
| 306 | string BenchmarkInfo() override { |
| 307 | return "Long string for suffix" ; |
| 308 | } |
| 309 | FINISH_BENCHMARK(StringSuffixLong) |
| 310 | |
| 311 | //----------------------- SUFFIX REGEX benchmark ----------------------------------// |
| 312 | DUCKDB_BENCHMARK(StringSuffixRegEX, "[string]" ) |
| 313 | STRING_DATA_GEN_BODY(4) |
| 314 | string GetQuery() override { |
| 315 | return "SELECT REGEXP_MATCHES(s1, '%a') FROM strings" ; |
| 316 | } |
| 317 | string BenchmarkInfo() override { |
| 318 | return "STRING suffix REGEX" ; |
| 319 | } |
| 320 | FINISH_BENCHMARK(StringSuffixRegEX) |
| 321 | |
| 322 | //----------------------- SUFFIX LIKE benchmark ----------------------------------// |
| 323 | DUCKDB_BENCHMARK(StringSuffixLikeShort, "[string]" ) |
| 324 | STRING_DATA_GEN_BODY(4) |
| 325 | string GetQuery() override { |
| 326 | return "SELECT s1 FROM strings WHERE s1 LIKE '%a'" ; |
| 327 | } |
| 328 | string BenchmarkInfo() override { |
| 329 | return "Short string for suffix with LIKE" ; |
| 330 | } |
| 331 | FINISH_BENCHMARK(StringSuffixLikeShort) |
| 332 | |
| 333 | DUCKDB_BENCHMARK(StringSuffixLikeLong, "[string]" ) |
| 334 | STRING_DATA_GEN_BODY(20) |
| 335 | string GetQuery() override { |
| 336 | return "SELECT s1 FROM strings WHERE s1 LIKE '%a'" ; |
| 337 | } |
| 338 | string BenchmarkInfo() override { |
| 339 | return "Long string for suffix with LIKE" ; |
| 340 | } |
| 341 | FINISH_BENCHMARK(StringSuffixLikeLong) |
| 342 | |