nevi-me commented on pull request #9329: URL: https://github.com/apache/arrow/pull/9329#issuecomment-791898261
Benchmarks on an ARM machine: ``` nevilledipale@Nevilles-MacBook-Air rust % critcmp master-bfa99d98e arcless-ef2250b9d -t 10 group arcless-ef2250b9d master-bfa99d98e ----- ----------------- ---------------- add_nulls_512 1.00 351.5±22.53ns ? B/sec 1.12 394.1±19.39ns ? B/sec array_from_vec 128 1.00 555.6±26.43ns ? B/sec 1.11 619.1±17.89ns ? B/sec array_slice 128 1.00 98.3±8.62ns ? B/sec 1.42 139.4±7.01ns ? B/sec array_slice 2048 1.00 105.8±13.50ns ? B/sec 1.37 145.1±8.93ns ? B/sec array_slice 512 1.00 101.9±6.16ns ? B/sec 1.41 143.3±9.44ns ? B/sec cast date32 to date64 512 1.00 246.7±17.81ns ? B/sec 1.23 303.7±14.81ns ? B/sec cast time32s to time32ms 512 1.00 223.1±13.25ns ? B/sec 1.18 262.3±16.68ns ? B/sec cast timestamp_ms to i64 512 1.00 98.6±9.32ns ? B/sec 1.56 153.7±7.79ns ? B/sec divide_scalar 512 1.00 189.9±12.11ns ? B/sec 1.16 220.4±8.05ns ? B/sec divide_scalar_nulls_512 1.00 186.6±14.60ns ? B/sec 1.21 225.1±11.87ns ? B/sec eq Float32 1.15 40.0±1.54µs ? B/sec 1.00 34.7±1.30µs ? B/sec equal_bool_512 1.00 25.8±1.34ns ? B/sec 1.12 28.9±1.15ns ? B/sec equal_bool_513 1.00 27.7±1.36ns ? B/sec 1.14 31.5±1.42ns ? B/sec gt Float32 1.15 40.6±2.00µs ? B/sec 1.00 35.2±1.75µs ? B/sec gt_eq Float32 1.20 42.1±2.70µs ? B/sec 1.00 35.2±1.60µs ? B/sec limit 512, 512 1.00 106.5±6.65ns ? B/sec 1.31 139.0±7.04ns ? B/sec lt Float32 1.15 40.5±1.97µs ? B/sec 1.00 35.3±1.73µs ? B/sec lt_eq Float32 1.14 40.3±2.40µs ? B/sec 1.00 35.2±1.50µs ? B/sec max nulls 512 1.00 889.7±59.93ns ? B/sec 3.26 2.9±0.11µs ? B/sec min nulls 512 1.00 916.4±70.85ns ? B/sec 2.64 2.4±0.10µs ? B/sec multiply 512 1.00 361.3±28.15ns ? B/sec 1.12 403.5±21.82ns ? B/sec neq Float32 1.16 40.6±1.72µs ? B/sec 1.00 34.9±1.68µs ? B/sec take bool 1024 1.00 4.2±0.30µs ? B/sec 1.29 5.4±0.29µs ? B/sec take bool 512 1.00 1750.4±100.33ns ? B/sec 1.38 2.4±0.19µs ? B/sec take bool nulls 1024 1.00 3.3±0.16µs ? B/sec 1.30 4.3±0.28µs ? B/sec take bool nulls 512 1.00 1635.3±119.87ns ? B/sec 1.20 1956.5±83.47ns ? B/sec take i32 512 1.00 421.9±11.67ns ? B/sec 1.14 480.0±27.45ns ? B/sec take str null values 1024 1.00 6.8±0.42µs ? B/sec 1.25 8.5±1.49µs ? B/sec ``` Only equality benches are worse, but not by that much. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org