apache
[improve](partial update) sort the rids to read for alignments to reduce the number of random accesses
Proposed changes
This PR sorts the rids to read for alignments in partial update to reduce the number of random accesses to improve performance.
Further comments
If this is a relatively large or complex change, kick off the discussion at [email protected] by explaining why you chose the solution you did and what alternatives you considered, etc...
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "Published by a pub.dev verified publisher"){kind=small}
(From new machine)TeamCity pipeline, clickbench performance test result: the sum of best hot time: 44.7 seconds stream load tsv: 576 seconds loaded 74807831229 Bytes, about 123 MB/s stream load json: 18 seconds loaded 2358488459 Bytes, about 124 MB/s stream load orc: 66 seconds loaded 1101869774 Bytes, about 15 MB/s stream load parquet: 32 seconds loaded 861443392 Bytes, about 25 MB/s insert into select: 29.6 seconds inserted 10000000 Rows, about 337K ops/s storage size: 17099463712 Bytes
TPC-H test result on machine: 'aliyun_ecs.c7a.8xlarge_32C64G'
Tpch sf100 test result on commit 5f70575bd909d27606ff4ee00d7ebb0cccd37e06, data reload: false
run tpch-sf100 query with default conf and session variables
q1 4914 4636 4639 4636
q2 364 143 159 143
q3 2036 1928 1830 1830
q4 1403 1262 1221 1221
q5 3949 3931 3987 3931
q6 252 130 130 130
q7 1450 895 867 867
q8 2815 2795 2791 2791
q9 9691 9838 9469 9469
q10 3450 3488 3501 3488
q11 375 254 250 250
q12 438 289 289 289
q13 4561 3769 3807 3769
q14 316 278 302 278
q15 573 518 520 518
q16 671 596 581 581
q17 1148 957 949 949
q18 7929 7570 7612 7570
q19 1670 1675 1696 1675
q20 526 308 305 305
q21 4453 4009 4085 4009
q22 477 375 360 360
Total cold run time: 53461 ms
Total hot run time: 49059 ms
run tpch-sf100 query with default conf and set session variable runtime_filter_mode=off
q1 4624 4572 4594 4572
q2 343 223 254 223
q3 4030 4000 4010 4000
q4 2731 2727 2718 2718
q5 9781 9780 9666 9666
q6 248 120 125 120
q7 3067 2494 2488 2488
q8 4426 4433 4414 4414
q9 12984 12871 12948 12871
q10 4039 4123 4151 4123
q11 873 729 656 656
q12 978 830 820 820
q13 4319 3551 3553 3551
q14 382 340 353 340
q15 567 510 520 510
q16 723 672 658 658
q17 3806 3865 3858 3858
q18 9764 9310 9236 9236
q19 1802 1782 1806 1782
q20 2411 2086 2071 2071
q21 8827 8585 8504 8504
q22 922 835 855 835
Total cold run time: 81647 ms
Total hot run time: 78016 ms
We're closing this PR because it hasn't been updated in a while. This isn't a judgement on the merit of the PR in any way. It's just a way of keeping the PR queue manageable. If you'd like to revive this PR, please reopen it and feel free a maintainer to remove the Stale tag!