perf(encoding): use estimate size in data chunk encoding #8924

Honeta · 2023-03-31T10:02:51Z

I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

as title

Checklist For Contributors

I have written necessary rustdoc comments
I have added necessary unit tests and integration tests
I have added fuzzing tests or opened an issue to track them. (Optional, recommended for new SQL features Sqlsmith: Sql feature generation #7934).
I have demonstrated that backward compatibility is not broken by breaking changes and created issues to track deprecated features to be removed in the future. (Please refer to the issue)
All checks passed in ./risedev check (or alias, ./risedev c)

Checklist For Reviewers

I have requested macro/micro-benchmarks as this PR can affect performance substantially, and the results are shown.

Documentation

My PR DOES NOT contain user-facing changes.

benchmark:

data chunk encoding: Int16, 128 rows, Pr[null]=0
                        time:   [5.0985 µs 5.0997 µs 5.1011 µs]
                        change: [-60.708% -60.585% -60.481%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  2 (2.00%) high mild
  3 (3.00%) high severe

data chunk encoding: Int16, 1024 rows, Pr[null]=0
                        time:   [70.243 µs 70.277 µs 70.334 µs]
                        change: [-45.606% -45.550% -45.464%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 8 outliers among 100 measurements (8.00%)
  2 (2.00%) high mild
  6 (6.00%) high severe

data chunk encoding: Int16, 128 rows, Pr[null]=0.01
                        time:   [9.0632 µs 9.0638 µs 9.0645 µs]
                        change: [-51.416% -51.347% -51.211%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 2 outliers among 100 measurements (2.00%)
  2 (2.00%) high severe

data chunk encoding: Int16, 1024 rows, Pr[null]=0.01
                        time:   [63.245 µs 63.248 µs 63.252 µs]
                        change: [-51.149% -51.129% -51.111%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  1 (1.00%) low severe
  1 (1.00%) low mild
  3 (3.00%) high mild
  4 (4.00%) high severe

data chunk encoding: Int16, 128 rows, Pr[null]=0.1
                        time:   [7.4565 µs 7.4573 µs 7.4582 µs]
                        change: [-53.708% -53.633% -53.504%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 8 outliers among 100 measurements (8.00%)
  1 (1.00%) low severe
  5 (5.00%) high mild
  2 (2.00%) high severe

data chunk encoding: Int16, 1024 rows, Pr[null]=0.1
                        time:   [59.937 µs 59.943 µs 59.950 µs]
                        change: [-53.571% -53.542% -53.502%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  4 (4.00%) high mild
  1 (1.00%) high severe

data chunk encoding: String, 128 rows, Pr[null]=0
                        time:   [11.026 µs 11.032 µs 11.044 µs]
                        change: [-30.353% -30.099% -29.848%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  1 (1.00%) high mild
  4 (4.00%) high severe

data chunk encoding: String, 1024 rows, Pr[null]=0
                        time:   [97.505 µs 97.516 µs 97.527 µs]
                        change: [-32.983% -32.899% -32.825%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  1 (1.00%) low mild
  3 (3.00%) high mild
  2 (2.00%) high severe

data chunk encoding: String, 128 rows, Pr[null]=0.01
                        time:   [10.976 µs 10.978 µs 10.980 µs]
                        change: [-24.373% -24.341% -24.295%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 10 outliers among 100 measurements (10.00%)
  2 (2.00%) low mild
  6 (6.00%) high mild
  2 (2.00%) high severe

data chunk encoding: String, 1024 rows, Pr[null]=0.01
                        time:   [96.907 µs 96.923 µs 96.944 µs]
                        change: [-34.879% -34.853% -34.832%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 7 outliers among 100 measurements (7.00%)
  1 (1.00%) low mild
  4 (4.00%) high mild
  2 (2.00%) high severe

data chunk encoding: String, 128 rows, Pr[null]=0.1
                        time:   [10.756 µs 10.758 µs 10.760 µs]
                        change: [-34.491% -34.474% -34.456%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  1 (1.00%) low mild
  2 (2.00%) high mild
  2 (2.00%) high severe

data chunk encoding: String, 1024 rows, Pr[null]=0.1
                        time:   [94.911 µs 94.924 µs 94.938 µs]
                        change: [-45.633% -45.613% -45.596%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) low severe
  1 (1.00%) low mild
  1 (1.00%) high mild
  1 (1.00%) high severe

data chunk encoding: Int16 and String, 128 rows, Pr[null]=0
                        time:   [10.961 µs 10.962 µs 10.964 µs]
                        change: [-34.081% -33.991% -33.764%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  1 (1.00%) low severe
  1 (1.00%) low mild
  2 (2.00%) high mild
  1 (1.00%) high severe

data chunk encoding: Int16 and String, 1024 rows, Pr[null]=0
                        time:   [107.45 µs 107.48 µs 107.52 µs]
                        change: [-25.952% -25.843% -25.672%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  2 (2.00%) high mild
  4 (4.00%) high severe

data chunk encoding: Int16 and String, 128 rows, Pr[null]=0.01
                        time:   [10.883 µs 10.885 µs 10.887 µs]
                        change: [-35.649% -35.442% -35.241%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 3 outliers among 100 measurements (3.00%)
  1 (1.00%) low mild
  2 (2.00%) high severe

data chunk encoding: Int16 and String, 1024 rows, Pr[null]=0.01
                        time:   [92.819 µs 92.829 µs 92.839 µs]
                        change: [-37.325% -37.120% -36.926%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 7 outliers among 100 measurements (7.00%)
  4 (4.00%) low mild
  2 (2.00%) high mild
  1 (1.00%) high severe

data chunk encoding: Int16 and String, 128 rows, Pr[null]=0.1
                        time:   [11.008 µs 11.009 µs 11.011 µs]
                        change: [-33.145% -33.052% -32.900%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 7 outliers among 100 measurements (7.00%)
  2 (2.00%) low mild
  2 (2.00%) high mild
  3 (3.00%) high severe

data chunk encoding: Int16 and String, 1024 rows, Pr[null]=0.1
                        time:   [94.356 µs 94.364 µs 94.372 µs]
                        change: [-35.836% -35.695% -35.611%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  3 (3.00%) low severe
  1 (1.00%) high mild
  2 (2.00%) high severe

data chunk encoding: Int16, Int32, Int64 and String, 128 rows, Pr[null]=0
                        time:   [13.827 µs 13.828 µs 13.830 µs]
                        change: [-60.049% -60.000% -59.879%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 8 outliers among 100 measurements (8.00%)
  1 (1.00%) low severe
  2 (2.00%) low mild
  2 (2.00%) high mild
  3 (3.00%) high severe

data chunk encoding: Int16, Int32, Int64 and String, 1024 rows, Pr[null]=0
                        time:   [131.10 µs 131.12 µs 131.13 µs]
                        change: [-52.724% -52.711% -52.697%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  1 (1.00%) low severe
  1 (1.00%) high mild
  3 (3.00%) high severe

data chunk encoding: Int16, Int32, Int64 and String, 128 rows, Pr[null]=0.01
                        time:   [13.837 µs 13.838 µs 13.841 µs]
                        change: [-59.884% -59.820% -59.723%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high severe

data chunk encoding: Int16, Int32, Int64 and String, 1024 rows, Pr[null]=0.01
                        time:   [130.70 µs 130.71 µs 130.73 µs]
                        change: [-53.020% -52.903% -52.827%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  1 (1.00%) low severe
  1 (1.00%) low mild
  4 (4.00%) high mild
  3 (3.00%) high severe

data chunk encoding: Int16, Int32, Int64 and String, 128 rows, Pr[null]=0.1
                        time:   [13.598 µs 13.599 µs 13.600 µs]
                        change: [-58.051% -57.943% -57.882%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 7 outliers among 100 measurements (7.00%)
  1 (1.00%) low severe
  2 (2.00%) low mild
  3 (3.00%) high mild
  1 (1.00%) high severe

data chunk encoding: Int16, Int32, Int64 and String, 1024 rows, Pr[null]=0.1
                        time:   [115.88 µs 115.90 µs 115.91 µs]
                        change: [-56.546% -56.537% -56.528%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 3 outliers among 100 measurements (3.00%)
  2 (2.00%) low mild
  1 (1.00%) high mild

codecov · 2023-03-31T10:25:33Z

Codecov Report

Merging #8924 (293c864) into main (ee6d44f) will increase coverage by 0.01%.
The diff coverage is 66.32%.

@@            Coverage Diff             @@
##             main    #8924      +/-   ##
==========================================
+ Coverage   70.90%   70.91%   +0.01%     
==========================================
  Files        1196     1196              
  Lines      198786   198881      +95     
==========================================
+ Hits       140940   141028      +88     
- Misses      57846    57853       +7

Flag	Coverage Δ
rust	`70.91% <66.32%> (+0.01%)`	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/common/src/array/data_chunk.rs	`89.22% <38.77%> (-4.32%)`	⬇️
src/common/src/util/value_encoding/mod.rs	`92.75% <93.87%> (+0.17%)`	⬆️

... and 6 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

src/common/src/array/data_chunk.rs

st1page

It could be more cache-friendly if we can use a vectorized addition. But we need alloc the sum array in advance. We need a trade-off between them maybe. But not emergent.

Co-authored-by: stonepage <[email protected]>

…nk_encoding

github-actions

license-eye has totally checked 3176 files.

Valid	Invalid	Ignored	Fixed
1477	2	1697	0

Click to see the invalid file list

src/common/benches/bench_data_chunk_encoding.rs
src/common/src/test_utils/rand_chunk.rs

src/common/benches/bench_data_chunk_encoding.rs

src/common/src/test_utils/rand_chunk.rs

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

src/common/src/util/value_encoding/mod.rs

use estimate size in data chunk encoding

4447078

Honeta requested a review from st1page March 31, 2023 10:02

github-actions bot added the type/perf label Mar 31, 2023

st1page reviewed Mar 31, 2023

View reviewed changes

src/common/src/array/data_chunk.rs Show resolved Hide resolved

src/common/src/array/data_chunk.rs Show resolved Hide resolved

st1page reviewed Mar 31, 2023

View reviewed changes

Apply suggestions from code review

9005952

Co-authored-by: stonepage <[email protected]>

st1page approved these changes Apr 3, 2023

View reviewed changes

Honeta added 2 commits April 3, 2023 07:41

Merge remote-tracking branch 'origin/main' into xinjing/perf_data_chu…

e8d3d9d

…nk_encoding

add bench

c6733af

github-actions bot reviewed Apr 3, 2023

View reviewed changes

src/common/benches/bench_data_chunk_encoding.rs Outdated Show resolved Hide resolved

src/common/src/test_utils/rand_chunk.rs Outdated Show resolved Hide resolved

Honeta and others added 13 commits April 3, 2023 12:13

try to improve locality

4420b5c

Update src/common/benches/bench_data_chunk_encoding.rs

6257986

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

Update src/common/src/test_utils/rand_chunk.rs

22fd207

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

se row_len_fixed and iter in col_variable

adacf62

add string only bench

fc16c44

remove some comment

c4aa8e8

remove redundant field names in struct initialization

f8e0cc8

Merge branch 'main' into xinjing/perf_data_chunk_encoding

cebbe43

use Vec instead of BytesMut for data chunk serialize

e55d55e

use with_capacity instead of reserve, capacity plus one

4980651

Merge branch 'main' into xinjing/perf_data_chunk_encoding

c437853

fix try_get_exact_serialize_datum_size

01382c5

try_get_exact_serialize_datum_size support more type

f2d0591

st1page reviewed Apr 10, 2023

View reviewed changes

src/common/src/util/value_encoding/mod.rs Outdated Show resolved Hide resolved

src/common/src/util/value_encoding/mod.rs Show resolved Hide resolved

Honeta added 5 commits April 10, 2023 08:52

use option map

c447df8

add ut for try_get_exact_serialize_datum_size

7dbb38e

improve code

1f04b11

remove useless vec![]

d74427f

addone more bench example

16d0e74

Honeta added 2 commits April 11, 2023 10:57

Merge branch 'main' into xinjing/perf_data_chunk_encoding

94dd827

fix empty buffer

293c864

st1page added this pull request to the merge queue Apr 11, 2023

Merged via the queue into main with commit bcc00c0 Apr 11, 2023

st1page deleted the xinjing/perf_data_chunk_encoding branch April 11, 2023 06:12

st1page mentioned this pull request Apr 11, 2023

perf(common): reduce buffer reallocations in HashKey::build #8885

Merged

7 tasks

kwannoel mentioned this pull request Apr 21, 2023

perf(hashkey): refine the memory reserving for SerializedKey #9305

Closed

kwannoel mentioned this pull request May 4, 2023

perf(hashkey): refine the memory reserving for SerializedKey #9556

Merged

7 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

perf(encoding): use estimate size in data chunk encoding #8924

perf(encoding): use estimate size in data chunk encoding #8924

Honeta commented Mar 31, 2023 •

edited

Loading

codecov bot commented Mar 31, 2023 •

edited

Loading

st1page left a comment

github-actions bot left a comment

perf(encoding): use estimate size in data chunk encoding #8924

perf(encoding): use estimate size in data chunk encoding #8924

Conversation

Honeta commented Mar 31, 2023 • edited Loading

What's changed and what's your intention?

Checklist For Contributors

Checklist For Reviewers

Documentation

codecov bot commented Mar 31, 2023 • edited Loading

Codecov Report

st1page left a comment

Choose a reason for hiding this comment

github-actions bot left a comment

Choose a reason for hiding this comment

Honeta commented Mar 31, 2023 •

edited

Loading

codecov bot commented Mar 31, 2023 •

edited

Loading