/
githubmirror
/
spark
Обзор
Документация
Войти
/
githubmirror
/
spark
Код
Запросы
0
Пакеты
0
Релизы
0
Аналитика
Безопасность
master
sql/core/benchmarks/AggregateBenchmark-results.txt
146 строк
11 KB
Dongjoon Hyun
[SPARK-58415][INFRA][TEST] Regenerate benchmark results
29 июл 2026, 10:51
29 июл 2026, 10:51
6b793ce
Код
Авторство
О чём код?
================================================================================================ aggregate without grouping ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor agg w/o group: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ agg w/o group wholestage off 34322 35781 2063 61.1 16.4 1.0X agg w/o group wholestage on 3881 3884 2 540.4 1.9 8.8X ================================================================================================ stat functions ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor stddev: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ stddev wholestage off 3645 3648 4 28.8 34.8 1.0X stddev wholestage on 1092 1098 4 96.0 10.4 3.3X OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor kurtosis: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ kurtosis wholestage off 20255 20293 54 5.2 193.2 1.0X kurtosis wholestage on 1112 1122 6 94.3 10.6 18.2X ================================================================================================ aggregate with linear keys ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor Aggregate w keys: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ codegen = F 6986 7075 125 12.0 83.3 1.0X codegen = T, hashmap = F 4643 4656 18 18.1 55.3 1.5X codegen = T, row-based hashmap = T 4636 4649 13 18.1 55.3 1.5X codegen = T, vectorized hashmap = T 4685 4726 36 17.9 55.8 1.5X ================================================================================================ aggregate with randomized keys ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor Aggregate w keys: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ codegen = F 7121 7145 34 11.8 84.9 1.0X codegen = T, hashmap = F 4569 4613 59 18.4 54.5 1.6X codegen = T, row-based hashmap = T 4567 4593 31 18.4 54.4 1.6X codegen = T, vectorized hashmap = T 4576 4589 22 18.3 54.6 1.6X ================================================================================================ aggregate with string key ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor Aggregate w string key: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ codegen = F 2360 2380 29 8.9 112.5 1.0X codegen = T, hashmap = F 1647 1653 9 12.7 78.5 1.4X codegen = T, row-based hashmap = T 1634 1653 34 12.8 77.9 1.4X codegen = T, vectorized hashmap = T 1637 1640 2 12.8 78.1 1.4X ================================================================================================ aggregate with decimal key ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor Aggregate w decimal key: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ codegen = F 2221 2267 64 9.4 105.9 1.0X codegen = T, hashmap = F 1488 1489 1 14.1 71.0 1.5X codegen = T, row-based hashmap = T 1506 1524 26 13.9 71.8 1.5X codegen = T, vectorized hashmap = T 1470 1501 44 14.3 70.1 1.5X ================================================================================================ aggregate with multiple key types ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor Aggregate w multiple keys: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ codegen = F 4212 4215 4 5.0 200.8 1.0X codegen = T, hashmap = F 2538 2571 47 8.3 121.0 1.7X codegen = T, row-based hashmap = T 1787 1791 6 11.7 85.2 2.4X codegen = T, vectorized hashmap = T 1715 1732 25 12.2 81.8 2.5X ================================================================================================ max function bytecode size of wholestagecodegen ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor max function bytecode size: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ codegen = F 345 358 8 1.9 525.9 1.0X codegen = T, hugeMethodLimit = 10000 134 152 13 4.9 204.3 2.6X codegen = T, hugeMethodLimit = 1500 129 141 10 5.1 197.5 2.7X ================================================================================================ cube ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor cube: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ cube wholestage off 2538 2552 20 2.1 484.0 1.0X cube wholestage on 1182 1194 12 4.4 225.5 2.1X ================================================================================================ hash and BytesToBytesMap ================================================================================================ OpenJDK 64-Bit Server VM 17.0.20+8-LTS on Linux 6.17.0-1020-azure AMD EPYC 9V74 80-Core Processor BytesToBytesMap: Best Time(ms) Avg Time(ms) Stdev(ms) Rate(M/s) Per Row(ns) Relative ------------------------------------------------------------------------------------------------------------------------ UnsafeRowhash 289 292 3 72.7 13.8 1.0X murmur3 hash 76 76 1 277.5 3.6 3.8X fast hash 47 47 2 450.6 2.2 6.2X arrayEqual 178 179 1 117.5 8.5 1.6X Java HashMap (Long) 65 69 4 323.5 3.1 4.5X Java HashMap (two ints) 86 89 3 244.6 4.1 3.4X Java HashMap (UnsafeRow) 570 573 3 36.8 27.2 0.5X LongToUnsafeRowMap (opt=false) 444 446 2 47.2 21.2 0.6X LongToUnsafeRowMap (opt=true) 92 96 5 227.2 4.4 3.1X BytesToBytesMap (off Heap) 535 550 10 39.2 25.5 0.5X BytesToBytesMap (on Heap) 527 529 2 39.8 25.1 0.5X Aggregate HashMap 35 35 0 607.4 1.6 8.4X