mapleFU commented on PR #40647: URL: https://github.com/apache/arrow/pull/40647#issuecomment-2017203489
On My MacOS M1Pro: Before: ``` TouchArea/size:4096/real_time/threads:1 10.9 ns 10.9 ns 57994989 bytes_per_second=350.334G/s items_per_second=91.838M/s TouchArea/size:4096/real_time/threads:2 5.60 ns 11.2 ns 124876054 bytes_per_second=681.267G/s items_per_second=178.59M/s TouchArea/size:4096/real_time/threads:4 2.86 ns 11.4 ns 245521852 bytes_per_second=1.30431T/s items_per_second=350.122M/s TouchArea/size:4096/real_time/threads:8 1.45 ns 11.6 ns 484530120 bytes_per_second=2.57092T/s items_per_second=690.125M/s TouchArea/size:4096/real_time/threads:16 1.38 ns 12.9 ns 590918384 bytes_per_second=2.69684T/s items_per_second=723.928M/s TouchArea/size:4096/real_time/threads:32 1.29 ns 12.8 ns 556564992 bytes_per_second=2.87679T/s items_per_second=772.233M/s TouchArea/size:65536/real_time/threads:1 208 ns 208 ns 3317775 bytes_per_second=293.541G/s items_per_second=4.80938M/s TouchArea/size:65536/real_time/threads:2 107 ns 213 ns 6526584 bytes_per_second=572.079G/s items_per_second=9.37295M/s TouchArea/size:65536/real_time/threads:4 54.1 ns 216 ns 12943336 bytes_per_second=1.10126T/s items_per_second=18.4761M/s TouchArea/size:65536/real_time/threads:8 27.5 ns 219 ns 25424256 bytes_per_second=2.16524T/s items_per_second=36.3268M/s TouchArea/size:65536/real_time/threads:16 28.7 ns 242 ns 29298224 bytes_per_second=2.07599T/s items_per_second=34.8294M/s TouchArea/size:65536/real_time/threads:32 25.1 ns 245 ns 38874976 bytes_per_second=2.37519T/s items_per_second=39.8491M/s TouchArea/size:1048576/real_time/threads:1 10319 ns 10315 ns 67168 bytes_per_second=94.6367G/s items_per_second=96.908k/s TouchArea/size:1048576/real_time/threads:2 5327 ns 10644 ns 133090 bytes_per_second=183.324G/s items_per_second=187.724k/s TouchArea/size:1048576/real_time/threads:4 2691 ns 10758 ns 261484 bytes_per_second=362.937G/s items_per_second=371.648k/s TouchArea/size:1048576/real_time/threads:8 1368 ns 10927 ns 506984 bytes_per_second=713.808G/s items_per_second=730.939k/s TouchArea/size:1048576/real_time/threads:16 1218 ns 11969 ns 600560 bytes_per_second=802.048G/s items_per_second=821.297k/s TouchArea/size:1048576/real_time/threads:32 1204 ns 12061 ns 557376 bytes_per_second=811.109G/s items_per_second=830.576k/s TouchArea/size:16777216/real_time/threads:1 253526 ns 253370 ns 2742 bytes_per_second=61.6306G/s items_per_second=3.94436k/s TouchArea/size:16777216/real_time/threads:2 139961 ns 279856 ns 4868 bytes_per_second=111.638G/s items_per_second=7.14484k/s TouchArea/size:16777216/real_time/threads:4 152334 ns 609277 ns 4424 bytes_per_second=102.571G/s items_per_second=6.56452k/s TouchArea/size:16777216/real_time/threads:8 153201 ns 1224218 ns 4512 bytes_per_second=101.99G/s items_per_second=6.52735k/s TouchArea/size:16777216/real_time/threads:16 137789 ns 1285625 ns 5296 bytes_per_second=113.398G/s items_per_second=7.25745k/s TouchArea/size:16777216/real_time/threads:32 126020 ns 1315402 ns 6016 bytes_per_second=123.988G/s items_per_second=7.93522k/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:1 32.2 ns 32.1 ns 21709751 items_per_second=31.0963M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:2 75.9 ns 152 ns 9743782 items_per_second=13.1729M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:4 103 ns 413 ns 7189620 items_per_second=9.67396M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:8 127 ns 1017 ns 5459352 items_per_second=7.85479M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:16 164 ns 1637 ns 4368576 items_per_second=6.08059M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:32 159 ns 1610 ns 3200000 items_per_second=6.27851M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:1 34.6 ns 34.6 ns 20213807 items_per_second=28.8995M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:2 78.6 ns 157 ns 8997418 items_per_second=12.7283M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:4 98.2 ns 393 ns 6788752 items_per_second=10.1842M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:8 130 ns 1037 ns 5417496 items_per_second=7.67272M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:16 167 ns 1659 ns 4431040 items_per_second=5.99549M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:32 161 ns 1639 ns 3200000 items_per_second=6.21225M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:1 34.5 ns 34.5 ns 20252808 items_per_second=28.9929M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:2 80.6 ns 161 ns 8622844 items_per_second=12.4024M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:4 101 ns 403 ns 7106452 items_per_second=9.91669M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:8 130 ns 1040 ns 5398432 items_per_second=7.66976M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:16 163 ns 1551 ns 4274144 items_per_second=6.12846M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:32 163 ns 1642 ns 3200000 items_per_second=6.15232M/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:1 212 ns 211 ns 3334565 items_per_second=4.71897M/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:2 743 ns 1418 ns 1107620 items_per_second=1.34635M/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:4 1094 ns 4374 ns 627532 items_per_second=914.037k/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:8 1769 ns 14099 ns 400296 items_per_second=565.15k/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:16 2024 ns 20213 ns 338096 items_per_second=494.161k/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:32 1931 ns 20111 ns 320000 items_per_second=517.831k/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:1 43.6 ns 43.6 ns 16031134 bytes_per_second=87.4222G/s items_per_second=22.9172M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:2 82.3 ns 164 ns 7117618 bytes_per_second=46.3634G/s items_per_second=12.1539M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:4 94.8 ns 379 ns 6668000 bytes_per_second=40.2382G/s items_per_second=10.5482M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:8 131 ns 1044 ns 5370112 bytes_per_second=29.0769G/s items_per_second=7.62233M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:16 163 ns 1620 ns 4179600 bytes_per_second=23.3966G/s items_per_second=6.13327M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:32 158 ns 1597 ns 3200000 bytes_per_second=24.2022G/s items_per_second=6.34447M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:1 259 ns 259 ns 2714421 bytes_per_second=235.671G/s items_per_second=3.86124M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:2 142 ns 284 ns 4417676 bytes_per_second=430.031G/s items_per_second=7.04563M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:4 130 ns 519 ns 4000000 bytes_per_second=468.521G/s items_per_second=7.67625M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:8 131 ns 1049 ns 5247904 bytes_per_second=464.59G/s items_per_second=7.61185M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:16 156 ns 1576 ns 4479440 bytes_per_second=391.971G/s items_per_second=6.42205M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:32 154 ns 1508 ns 4565600 bytes_per_second=397.094G/s items_per_second=6.50599M/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:1 10273 ns 10270 ns 68181 bytes_per_second=95.0594G/s items_per_second=97.3409k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:2 5309 ns 10616 ns 132128 bytes_per_second=183.938G/s items_per_second=188.353k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:4 2739 ns 10948 ns 255152 bytes_per_second=356.54G/s items_per_second=365.097k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:8 1511 ns 11467 ns 497368 bytes_per_second=646.375G/s items_per_second=661.888k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:16 1220 ns 12373 ns 609424 bytes_per_second=800.314G/s items_per_second=819.522k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:32 1223 ns 12554 ns 587936 bytes_per_second=798.588G/s items_per_second=817.754k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:1 1325250 ns 1323670 ns 528 bytes_per_second=11.7902G/s items_per_second=754.574/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:2 689854 ns 1379529 ns 988 bytes_per_second=22.6497G/s items_per_second=1.44958k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:4 519968 ns 2079136 ns 1320 bytes_per_second=30.0499G/s items_per_second=1.9232k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:8 155025 ns 1238565 ns 4536 bytes_per_second=100.79G/s items_per_second=6.45058k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:16 179778 ns 1879703 ns 3872 bytes_per_second=86.9129G/s items_per_second=5.56242k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:32 253345 ns 2907238 ns 2048 bytes_per_second=61.6748G/s items_per_second=3.94719k/s ``` After: ``` TouchArea/size:4096/real_time/threads:1 10.9 ns 10.9 ns 58233346 bytes_per_second=349.422G/s items_per_second=91.5988M/s TouchArea/size:4096/real_time/threads:2 5.60 ns 11.2 ns 125193588 bytes_per_second=681.472G/s items_per_second=178.644M/s TouchArea/size:4096/real_time/threads:4 2.85 ns 11.4 ns 245458448 bytes_per_second=1.3062T/s items_per_second=350.63M/s TouchArea/size:4096/real_time/threads:8 1.48 ns 11.8 ns 484327440 bytes_per_second=2.52042T/s items_per_second=676.57M/s TouchArea/size:4096/real_time/threads:16 1.30 ns 13.0 ns 583730720 bytes_per_second=2.86103T/s items_per_second=768.003M/s TouchArea/size:4096/real_time/threads:32 1.39 ns 13.0 ns 572841536 bytes_per_second=2.67056T/s items_per_second=716.874M/s TouchArea/size:65536/real_time/threads:1 207 ns 206 ns 3384914 bytes_per_second=295.514G/s items_per_second=4.8417M/s TouchArea/size:65536/real_time/threads:2 106 ns 212 ns 6590914 bytes_per_second=574.778G/s items_per_second=9.41716M/s TouchArea/size:65536/real_time/threads:4 54.2 ns 217 ns 12962756 bytes_per_second=1126.11G/s items_per_second=18.4502M/s TouchArea/size:65536/real_time/threads:8 27.4 ns 219 ns 25492952 bytes_per_second=2.17293T/s items_per_second=36.4557M/s TouchArea/size:65536/real_time/threads:16 24.6 ns 247 ns 28787872 bytes_per_second=2.42679T/s items_per_second=40.7148M/s TouchArea/size:65536/real_time/threads:32 24.6 ns 247 ns 33010080 bytes_per_second=2.4265T/s items_per_second=40.7098M/s TouchArea/size:1048576/real_time/threads:1 10282 ns 10277 ns 67234 bytes_per_second=94.9769G/s items_per_second=97.2564k/s TouchArea/size:1048576/real_time/threads:2 5324 ns 10643 ns 131616 bytes_per_second=183.436G/s items_per_second=187.838k/s TouchArea/size:1048576/real_time/threads:4 2686 ns 10738 ns 255752 bytes_per_second=363.547G/s items_per_second=372.272k/s TouchArea/size:1048576/real_time/threads:8 1375 ns 10961 ns 507424 bytes_per_second=710.368G/s items_per_second=727.417k/s TouchArea/size:1048576/real_time/threads:16 1210 ns 12051 ns 615152 bytes_per_second=807.181G/s items_per_second=826.553k/s TouchArea/size:1048576/real_time/threads:32 1190 ns 12141 ns 604992 bytes_per_second=820.868G/s items_per_second=840.569k/s TouchArea/size:16777216/real_time/threads:1 273083 ns 273033 ns 2702 bytes_per_second=57.217G/s items_per_second=3.66189k/s TouchArea/size:16777216/real_time/threads:2 141548 ns 283059 ns 4964 bytes_per_second=110.387G/s items_per_second=7.06475k/s TouchArea/size:16777216/real_time/threads:4 158891 ns 632653 ns 4412 bytes_per_second=98.3379G/s items_per_second=6.29363k/s TouchArea/size:16777216/real_time/threads:8 155452 ns 1216134 ns 4608 bytes_per_second=100.513G/s items_per_second=6.43284k/s TouchArea/size:16777216/real_time/threads:16 134444 ns 1315938 ns 5296 bytes_per_second=116.219G/s items_per_second=7.43803k/s TouchArea/size:16777216/real_time/threads:32 127480 ns 1309385 ns 5472 bytes_per_second=122.569G/s items_per_second=7.84439k/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:1 32.1 ns 32.1 ns 21690403 items_per_second=31.1574M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:2 82.7 ns 165 ns 7816064 items_per_second=12.091M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:4 140 ns 560 ns 4934076 items_per_second=7.14323M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:8 173 ns 1384 ns 4104304 items_per_second=5.77505M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:16 286 ns 2883 ns 2458128 items_per_second=3.49514M/s AllocateDeallocate<SystemAlloc>/size:4096/real_time/threads:32 285 ns 2870 ns 2917728 items_per_second=3.51277M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:1 34.6 ns 34.6 ns 20263456 items_per_second=28.9233M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:2 90.9 ns 182 ns 7188236 items_per_second=11.0032M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:4 149 ns 597 ns 4807920 items_per_second=6.69232M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:8 171 ns 1365 ns 4155448 items_per_second=5.85232M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:16 283 ns 2822 ns 2480400 items_per_second=3.53548M/s AllocateDeallocate<SystemAlloc>/size:65536/real_time/threads:32 283 ns 2905 ns 3027264 items_per_second=3.53694M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:1 34.5 ns 34.4 ns 20251705 items_per_second=29.0154M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:2 87.2 ns 174 ns 7039696 items_per_second=11.4744M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:4 163 ns 651 ns 5453856 items_per_second=6.14621M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:8 179 ns 1419 ns 4187808 items_per_second=5.58798M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:16 281 ns 2866 ns 2439200 items_per_second=3.56015M/s AllocateDeallocate<SystemAlloc>/size:1048576/real_time/threads:32 283 ns 2907 ns 2820320 items_per_second=3.53071M/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:1 210 ns 210 ns 3335281 items_per_second=4.76775M/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:2 595 ns 1189 ns 1236228 items_per_second=1.68083M/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:4 1145 ns 4575 ns 663448 items_per_second=873.578k/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:8 1756 ns 13963 ns 405592 items_per_second=569.538k/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:16 2006 ns 20012 ns 351120 items_per_second=498.43k/s AllocateDeallocate<SystemAlloc>/size:16777216/real_time/threads:32 1872 ns 19736 ns 446720 items_per_second=534.2k/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:1 43.3 ns 43.3 ns 16139653 bytes_per_second=88.1665G/s items_per_second=23.1123M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:2 83.6 ns 167 ns 6850320 bytes_per_second=45.647G/s items_per_second=11.9661M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:4 146 ns 583 ns 5593792 bytes_per_second=26.1403G/s items_per_second=6.85252M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:8 194 ns 1549 ns 3623832 bytes_per_second=19.6813G/s items_per_second=5.15933M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:16 288 ns 2903 ns 2478016 bytes_per_second=13.2574G/s items_per_second=3.47535M/s AllocateTouchDeallocate<SystemAlloc>/size:4096/real_time/threads:32 284 ns 2891 ns 2915552 bytes_per_second=13.4396G/s items_per_second=3.52311M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:1 256 ns 256 ns 2729001 bytes_per_second=238.069G/s items_per_second=3.90053M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:2 134 ns 267 ns 5425282 bytes_per_second=456.636G/s items_per_second=7.48152M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:4 130 ns 519 ns 5629480 bytes_per_second=470.541G/s items_per_second=7.70935M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:8 207 ns 1609 ns 3285216 bytes_per_second=294.611G/s items_per_second=4.8269M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:16 262 ns 2341 ns 2897648 bytes_per_second=233.093G/s items_per_second=3.819M/s AllocateTouchDeallocate<SystemAlloc>/size:65536/real_time/threads:32 240 ns 2212 ns 2858336 bytes_per_second=254.517G/s items_per_second=4.17001M/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:1 10957 ns 10947 ns 64417 bytes_per_second=89.1241G/s items_per_second=91.2631k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:2 5524 ns 11038 ns 127042 bytes_per_second=176.772G/s items_per_second=181.015k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:4 2763 ns 11041 ns 200264 bytes_per_second=353.498G/s items_per_second=361.982k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:8 1477 ns 11775 ns 459144 bytes_per_second=661.092G/s items_per_second=676.958k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:16 1298 ns 12362 ns 588272 bytes_per_second=752.097G/s items_per_second=770.147k/s AllocateTouchDeallocate<SystemAlloc>/size:1048576/real_time/threads:32 1246 ns 12478 ns 583680 bytes_per_second=784.035G/s items_per_second=802.852k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:1 1320591 ns 1320194 ns 520 bytes_per_second=11.8318G/s items_per_second=757.237/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:2 703512 ns 1406840 ns 1010 bytes_per_second=22.21G/s items_per_second=1.42144k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:4 555389 ns 2220354 ns 1120 bytes_per_second=28.1335G/s items_per_second=1.80054k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:8 401878 ns 3200275 ns 1744 bytes_per_second=38.88G/s items_per_second=2.48832k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:16 158491 ns 1588349 ns 4128 bytes_per_second=98.5858G/s items_per_second=6.30949k/s AllocateTouchDeallocate<SystemAlloc>/size:16777216/real_time/threads:32 259015 ns 2796938 ns 3232 bytes_per_second=60.3246G/s items_per_second=3.86078k/s ``` -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
