You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@arrow.apache.org by "Dandandan (via GitHub)" <gi...@apache.org> on 2026/02/14 19:28:20 UTC
[PR] [PoC] Selectivity threshold [arrow-rs]
Dandandan opened a new pull request, #9414:
URL: https://github.com/apache/arrow-rs/pull/9414
# Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax.
-->
- Closes #NNN.
# Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes.
-->
# What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR.
-->
# Are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example, are they covered by existing tests)?
-->
# Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
If there are any breaking changes to public APIs, please call them out.
-->
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3916037499
Nice
```
arrow_reader_clickbench/async/Q36 1.08 28.6±0.35ms ? ?/sec 1.00 26.6±0.34ms ? ?/sec
arrow_reader_clickbench/async/Q37 1.28 9.5±0.11ms ? ?/sec 1.00 7.4±0.17ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.00 24.7±0.55ms ? ?/sec 1.00 24.7±0.25ms ? ?/sec
arrow_reader_clickbench/async/Q40 1.37 13.2±0.17ms ? ?/sec 1.00 9.7±0.13ms ? ?/sec
arrow_reader_clickbench/async/Q41 1.31 10.3±0.14ms ? ?/sec 1.00 7.9±0.11ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.06 5.8±0.05ms ? ?/sec 1.00 5.4±0.04ms ? ?/sec```
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096223199
🤖 Arrow criterion benchmark completed (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096060759)
<details><summary>Details</summary>
<p>
```
group main selectivity_threshold
----- ---- ---------------------
arrow_reader_clickbench/async/Q1 1.00 1101.9±6.39µs ? ?/sec 1.01 1107.6±8.29µs ? ?/sec
arrow_reader_clickbench/async/Q10 1.00 6.7±0.04ms ? ?/sec 1.00 6.7±0.04ms ? ?/sec
arrow_reader_clickbench/async/Q11 1.00 7.7±0.07ms ? ?/sec 1.00 7.7±0.05ms ? ?/sec
arrow_reader_clickbench/async/Q12 1.00 14.3±0.05ms ? ?/sec 1.01 14.4±0.12ms ? ?/sec
arrow_reader_clickbench/async/Q13 1.00 17.1±0.07ms ? ?/sec 1.00 17.1±0.09ms ? ?/sec
arrow_reader_clickbench/async/Q14 1.00 15.9±0.06ms ? ?/sec 1.01 16.0±0.04ms ? ?/sec
arrow_reader_clickbench/async/Q19 1.00 3.1±0.03ms ? ?/sec 1.02 3.1±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q20 1.00 76.9±4.18ms ? ?/sec 1.10 84.4±9.36ms ? ?/sec
arrow_reader_clickbench/async/Q21 1.00 95.9±0.69ms ? ?/sec 1.07 102.5±5.83ms ? ?/sec
arrow_reader_clickbench/async/Q22 1.00 130.2±4.15ms ? ?/sec 1.02 132.3±0.61ms ? ?/sec
arrow_reader_clickbench/async/Q23 1.01 241.9±0.71ms ? ?/sec 1.00 239.4±0.77ms ? ?/sec
arrow_reader_clickbench/async/Q24 1.00 19.3±0.07ms ? ?/sec 1.00 19.4±0.08ms ? ?/sec
arrow_reader_clickbench/async/Q27 1.03 58.4±0.26ms ? ?/sec 1.00 56.7±0.19ms ? ?/sec
arrow_reader_clickbench/async/Q28 1.00 57.1±0.19ms ? ?/sec 1.00 57.0±0.17ms ? ?/sec
arrow_reader_clickbench/async/Q30 1.00 18.6±0.05ms ? ?/sec 1.00 18.6±0.07ms ? ?/sec
arrow_reader_clickbench/async/Q36 1.00 14.8±0.24ms ? ?/sec 1.00 14.7±0.13ms ? ?/sec
arrow_reader_clickbench/async/Q37 1.00 5.5±0.03ms ? ?/sec 1.02 5.6±0.02ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.01 13.0±0.14ms ? ?/sec 1.00 12.9±0.11ms ? ?/sec
arrow_reader_clickbench/async/Q39 1.00 23.8±0.21ms ? ?/sec 1.00 23.8±0.19ms ? ?/sec
arrow_reader_clickbench/async/Q40 1.00 5.8±0.03ms ? ?/sec 1.00 5.8±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q41 1.00 5.1±0.02ms ? ?/sec 1.03 5.2±0.02ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.00 3.6±0.02ms ? ?/sec 1.00 3.6±0.01ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q1 1.00 1081.2±4.84µs ? ?/sec 1.01 1093.5±14.06µs ? ?/sec
arrow_reader_clickbench/async_object_store/Q10 1.00 6.5±0.03ms ? ?/sec 1.02 6.6±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q11 1.00 7.5±0.04ms ? ?/sec 1.00 7.5±0.05ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q12 1.00 14.2±0.04ms ? ?/sec 1.01 14.3±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q13 1.00 16.8±0.07ms ? ?/sec 1.00 16.8±0.07ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q14 1.00 15.8±0.04ms ? ?/sec 1.00 15.8±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q19 1.00 3.0±0.02ms ? ?/sec 1.00 3.0±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q20 1.02 72.0±0.54ms ? ?/sec 1.00 70.8±0.29ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q21 1.02 80.8±0.48ms ? ?/sec 1.00 79.3±0.22ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q22 1.03 99.7±0.73ms ? ?/sec 1.00 96.6±0.29ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q23 1.03 221.4±0.41ms ? ?/sec 1.00 215.4±1.56ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q24 1.00 19.0±0.10ms ? ?/sec 1.01 19.2±0.13ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q27 1.02 56.9±0.48ms ? ?/sec 1.00 55.9±0.29ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q28 1.00 56.3±0.18ms ? ?/sec 1.00 56.1±0.24ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q30 1.00 18.3±0.05ms ? ?/sec 1.01 18.4±0.08ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q36 1.01 14.2±0.22ms ? ?/sec 1.00 14.1±0.22ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q37 1.00 5.4±0.01ms ? ?/sec 1.01 5.4±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q38 1.00 12.5±0.18ms ? ?/sec 1.00 12.4±0.23ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q39 1.00 22.5±0.17ms ? ?/sec 1.01 22.7±0.25ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q40 1.00 5.5±0.03ms ? ?/sec 1.01 5.6±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q41 1.00 4.9±0.02ms ? ?/sec 1.03 5.0±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q42 1.00 3.4±0.01ms ? ?/sec 1.01 3.5±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q1 1.00 872.5±5.37µs ? ?/sec 1.01 879.4±1.75µs ? ?/sec
arrow_reader_clickbench/sync/Q10 1.00 5.2±0.04ms ? ?/sec 1.00 5.1±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q11 1.00 6.1±0.02ms ? ?/sec 1.01 6.2±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q12 1.02 22.1±0.60ms ? ?/sec 1.00 21.7±0.12ms ? ?/sec
arrow_reader_clickbench/sync/Q13 1.00 27.9±0.75ms ? ?/sec 1.05 29.4±0.14ms ? ?/sec
arrow_reader_clickbench/sync/Q14 1.00 23.0±0.06ms ? ?/sec 1.15 26.6±0.35ms ? ?/sec
arrow_reader_clickbench/sync/Q19 1.01 2.7±0.02ms ? ?/sec 1.00 2.7±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q20 1.02 122.6±0.46ms ? ?/sec 1.00 119.8±0.37ms ? ?/sec
arrow_reader_clickbench/sync/Q21 1.02 98.0±0.35ms ? ?/sec 1.00 95.9±0.22ms ? ?/sec
arrow_reader_clickbench/sync/Q22 1.01 142.5±0.50ms ? ?/sec 1.00 141.6±0.64ms ? ?/sec
arrow_reader_clickbench/sync/Q23 1.00 276.3±13.33ms ? ?/sec 1.00 277.0±14.52ms ? ?/sec
arrow_reader_clickbench/sync/Q24 1.00 27.0±0.06ms ? ?/sec 1.01 27.1±0.15ms ? ?/sec
arrow_reader_clickbench/sync/Q27 1.02 107.6±0.46ms ? ?/sec 1.00 105.5±0.62ms ? ?/sec
arrow_reader_clickbench/sync/Q28 1.01 103.4±0.38ms ? ?/sec 1.00 102.4±0.29ms ? ?/sec
arrow_reader_clickbench/sync/Q30 1.01 18.7±0.05ms ? ?/sec 1.00 18.6±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q36 1.00 21.9±0.05ms ? ?/sec 1.00 22.0±0.04ms ? ?/sec
arrow_reader_clickbench/sync/Q37 1.00 6.9±0.04ms ? ?/sec 1.01 7.0±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q38 1.00 11.2±0.04ms ? ?/sec 1.00 11.2±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q39 1.00 20.5±0.06ms ? ?/sec 1.01 20.7±0.23ms ? ?/sec
arrow_reader_clickbench/sync/Q40 1.00 5.2±0.02ms ? ?/sec 1.01 5.3±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q41 1.00 5.6±0.04ms ? ?/sec 1.01 5.7±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q42 1.00 4.3±0.02ms ? ?/sec 1.01 4.4±0.02ms ? ?/sec
```
</p>
</details>
<details><summary>Resource Usage</summary>
**base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 779.4s |
| Peak memory | 3.1 GiB |
| Avg memory | 2.9 GiB |
| CPU user | 710.1s |
| CPU sys | 69.2s |
| Disk read | 0 B |
| Disk write | 2.0 GiB |
**branch**
| Metric | Value |
|--------|-------|
| Wall time | 782.1s |
| Peak memory | 3.2 GiB |
| Avg memory | 3.1 GiB |
| CPU user | 711.4s |
| CPU sys | 70.2s |
| Disk read | 0 B |
| Disk write | 171.4 MiB |
</details>
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096152206
🤖 Arrow criterion benchmark completed (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095977120)
<details><summary>Details</summary>
<p>
```
group main selectivity_threshold
----- ---- ---------------------
arrow_reader_clickbench/async/Q1 1.00 1081.2±5.61µs ? ?/sec 1.00 1077.7±3.29µs ? ?/sec
arrow_reader_clickbench/async/Q10 1.00 6.6±0.03ms ? ?/sec 1.00 6.5±0.02ms ? ?/sec
arrow_reader_clickbench/async/Q11 1.01 7.6±0.05ms ? ?/sec 1.00 7.6±0.04ms ? ?/sec
arrow_reader_clickbench/async/Q12 1.01 14.3±0.04ms ? ?/sec 1.00 14.1±0.04ms ? ?/sec
arrow_reader_clickbench/async/Q13 1.01 16.8±0.06ms ? ?/sec 1.00 16.7±0.08ms ? ?/sec
arrow_reader_clickbench/async/Q14 1.01 15.8±0.04ms ? ?/sec 1.00 15.7±0.05ms ? ?/sec
arrow_reader_clickbench/async/Q19 1.00 3.0±0.02ms ? ?/sec 1.01 3.1±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q20 1.18 84.1±9.60ms ? ?/sec 1.00 71.5±0.15ms ? ?/sec
arrow_reader_clickbench/async/Q21 1.27 106.1±2.62ms ? ?/sec 1.00 83.3±0.16ms ? ?/sec
arrow_reader_clickbench/async/Q22 1.00 134.2±5.81ms ? ?/sec 1.18 158.4±1.65ms ? ?/sec
arrow_reader_clickbench/async/Q23 1.03 246.7±2.46ms ? ?/sec 1.00 240.0±1.32ms ? ?/sec
arrow_reader_clickbench/async/Q24 1.01 19.2±0.09ms ? ?/sec 1.00 19.0±0.06ms ? ?/sec
arrow_reader_clickbench/async/Q27 1.02 58.1±0.22ms ? ?/sec 1.00 56.7±0.18ms ? ?/sec
arrow_reader_clickbench/async/Q28 1.02 58.1±0.31ms ? ?/sec 1.00 56.9±0.15ms ? ?/sec
arrow_reader_clickbench/async/Q30 1.01 18.5±0.07ms ? ?/sec 1.00 18.3±0.07ms ? ?/sec
arrow_reader_clickbench/async/Q36 1.12 15.1±0.20ms ? ?/sec 1.00 13.4±0.10ms ? ?/sec
arrow_reader_clickbench/async/Q37 1.19 5.4±0.02ms ? ?/sec 1.00 4.6±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.08 13.3±0.19ms ? ?/sec 1.00 12.3±0.11ms ? ?/sec
arrow_reader_clickbench/async/Q39 1.05 24.6±0.33ms ? ?/sec 1.00 23.4±0.21ms ? ?/sec
arrow_reader_clickbench/async/Q40 1.12 5.7±0.03ms ? ?/sec 1.00 5.1±0.02ms ? ?/sec
arrow_reader_clickbench/async/Q41 1.11 5.0±0.02ms ? ?/sec 1.00 4.5±0.04ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.08 3.6±0.02ms ? ?/sec 1.00 3.3±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q1 1.00 1065.7±2.65µs ? ?/sec 1.00 1063.1±6.31µs ? ?/sec
arrow_reader_clickbench/async_object_store/Q10 1.00 6.5±0.02ms ? ?/sec 1.01 6.5±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q11 1.00 7.4±0.03ms ? ?/sec 1.01 7.5±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q12 1.01 14.2±0.15ms ? ?/sec 1.00 14.1±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q13 1.02 16.8±0.14ms ? ?/sec 1.00 16.5±0.07ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q14 1.01 15.7±0.16ms ? ?/sec 1.00 15.6±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q19 1.00 3.0±0.02ms ? ?/sec 1.00 3.0±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q20 1.01 71.2±0.42ms ? ?/sec 1.00 70.5±0.21ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q21 1.00 80.4±0.47ms ? ?/sec 1.02 82.2±0.18ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q22 1.00 97.4±0.56ms ? ?/sec 1.53 149.2±0.22ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q23 1.02 229.4±0.49ms ? ?/sec 1.00 225.6±0.52ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q24 1.01 19.0±0.16ms ? ?/sec 1.00 18.8±0.05ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q27 1.01 56.4±0.36ms ? ?/sec 1.00 55.8±0.23ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q28 1.01 56.6±0.45ms ? ?/sec 1.00 56.0±0.17ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q30 1.01 18.3±0.14ms ? ?/sec 1.00 18.1±0.06ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q36 1.10 14.2±0.19ms ? ?/sec 1.00 12.9±0.11ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q37 1.17 5.4±0.03ms ? ?/sec 1.00 4.6±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q38 1.07 12.5±0.17ms ? ?/sec 1.00 11.6±0.10ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q39 1.04 23.2±0.47ms ? ?/sec 1.00 22.3±0.19ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q40 1.12 5.5±0.06ms ? ?/sec 1.00 4.9±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q41 1.11 4.8±0.06ms ? ?/sec 1.00 4.3±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q42 1.07 3.4±0.03ms ? ?/sec 1.00 3.2±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q1 1.01 878.2±12.14µs ? ?/sec 1.00 869.6±2.50µs ? ?/sec
arrow_reader_clickbench/sync/Q10 1.01 5.1±0.04ms ? ?/sec 1.00 5.1±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q11 1.01 6.1±0.10ms ? ?/sec 1.00 6.1±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q12 1.02 22.0±0.59ms ? ?/sec 1.00 21.5±0.12ms ? ?/sec
arrow_reader_clickbench/sync/Q13 1.00 27.8±0.73ms ? ?/sec 1.05 29.3±0.11ms ? ?/sec
arrow_reader_clickbench/sync/Q14 1.00 23.3±0.45ms ? ?/sec 1.14 26.6±0.35ms ? ?/sec
arrow_reader_clickbench/sync/Q19 1.02 2.7±0.02ms ? ?/sec 1.00 2.7±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q20 1.00 120.6±0.50ms ? ?/sec 1.00 120.5±0.32ms ? ?/sec
arrow_reader_clickbench/sync/Q21 1.00 97.5±2.16ms ? ?/sec 1.02 99.3±0.40ms ? ?/sec
arrow_reader_clickbench/sync/Q22 1.00 141.3±0.58ms ? ?/sec 1.36 192.6±0.71ms ? ?/sec
arrow_reader_clickbench/sync/Q23 1.04 280.2±14.45ms ? ?/sec 1.00 269.6±12.92ms ? ?/sec
arrow_reader_clickbench/sync/Q24 1.00 26.9±0.07ms ? ?/sec 1.00 26.8±0.06ms ? ?/sec
arrow_reader_clickbench/sync/Q27 1.01 106.6±0.41ms ? ?/sec 1.00 105.4±0.30ms ? ?/sec
arrow_reader_clickbench/sync/Q28 1.01 104.0±0.43ms ? ?/sec 1.00 102.8±0.35ms ? ?/sec
arrow_reader_clickbench/sync/Q30 1.00 18.6±0.06ms ? ?/sec 1.00 18.6±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q36 1.06 21.9±0.04ms ? ?/sec 1.00 20.6±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q37 1.23 6.9±0.01ms ? ?/sec 1.00 5.6±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q38 1.06 11.2±0.03ms ? ?/sec 1.00 10.5±0.04ms ? ?/sec
arrow_reader_clickbench/sync/Q39 1.00 20.5±0.08ms ? ?/sec 1.00 20.4±0.06ms ? ?/sec
arrow_reader_clickbench/sync/Q40 1.10 5.2±0.02ms ? ?/sec 1.00 4.7±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q41 1.38 5.7±0.02ms ? ?/sec 1.00 4.1±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q42 1.05 4.3±0.03ms ? ?/sec 1.00 4.2±0.02ms ? ?/sec
```
</p>
</details>
<details><summary>Resource Usage</summary>
**base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 783.1s |
| Peak memory | 3.1 GiB |
| Avg memory | 2.9 GiB |
| CPU user | 709.5s |
| CPU sys | 73.4s |
| Disk read | 0 B |
| Disk write | 1.9 GiB |
**branch**
| Metric | Value |
|--------|-------|
| Wall time | 785.6s |
| Peak memory | 3.2 GiB |
| Avg memory | 3.1 GiB |
| CPU user | 726.1s |
| CPU sys | 59.6s |
| Disk read | 0 B |
| Disk write | 171.4 MiB |
</details>
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on code in PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#discussion_r3006281264
##########
parquet/src/arrow/arrow_reader/mod.rs:
##########
@@ -430,6 +434,32 @@ impl<T> ArrowReaderBuilder<T> {
..self
}
}
+
+ /// Set a scatter threshold for filter deferral.
Review Comment:
Makes sense!
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095685694
🤖 Arrow criterion benchmark completed (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095543573)
<details><summary>Details</summary>
<p>
```
group main selectivity_threshold
----- ---- ---------------------
arrow_reader_clickbench/async/Q1 1.00 1092.8±13.79µs ? ?/sec 1.00 1090.9±4.96µs ? ?/sec
arrow_reader_clickbench/async/Q10 1.00 6.8±0.18ms ? ?/sec 1.01 6.9±0.08ms ? ?/sec
arrow_reader_clickbench/async/Q11 1.00 7.8±0.03ms ? ?/sec 1.00 7.7±0.05ms ? ?/sec
arrow_reader_clickbench/async/Q12 1.00 14.8±0.25ms ? ?/sec 1.00 14.9±0.21ms ? ?/sec
arrow_reader_clickbench/async/Q13 1.01 17.6±0.24ms ? ?/sec 1.00 17.5±0.34ms ? ?/sec
arrow_reader_clickbench/async/Q14 1.01 16.3±0.28ms ? ?/sec 1.00 16.1±0.20ms ? ?/sec
arrow_reader_clickbench/async/Q19 1.01 3.1±0.05ms ? ?/sec 1.00 3.1±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q20 1.09 78.1±4.42ms ? ?/sec 1.00 71.9±0.33ms ? ?/sec
arrow_reader_clickbench/async/Q21 1.21 96.9±0.94ms ? ?/sec 1.00 80.2±0.26ms ? ?/sec
arrow_reader_clickbench/async/Q22 1.17 131.5±5.71ms ? ?/sec 1.00 112.9±0.57ms ? ?/sec
arrow_reader_clickbench/async/Q23 1.03 247.5±1.51ms ? ?/sec 1.00 239.4±1.46ms ? ?/sec
arrow_reader_clickbench/async/Q24 1.02 19.9±0.25ms ? ?/sec 1.00 19.5±0.18ms ? ?/sec
arrow_reader_clickbench/async/Q27 1.02 58.4±0.59ms ? ?/sec 1.00 57.3±0.33ms ? ?/sec
arrow_reader_clickbench/async/Q28 1.01 58.2±0.58ms ? ?/sec 1.00 57.5±0.33ms ? ?/sec
arrow_reader_clickbench/async/Q30 1.01 18.9±0.16ms ? ?/sec 1.00 18.8±0.19ms ? ?/sec
arrow_reader_clickbench/async/Q36 1.07 16.0±0.45ms ? ?/sec 1.00 14.9±0.40ms ? ?/sec
arrow_reader_clickbench/async/Q37 1.03 5.5±0.03ms ? ?/sec 1.00 5.3±0.05ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.08 14.2±0.41ms ? ?/sec 1.00 13.2±0.18ms ? ?/sec
arrow_reader_clickbench/async/Q39 1.04 25.1±0.60ms ? ?/sec 1.00 24.1±0.34ms ? ?/sec
arrow_reader_clickbench/async/Q40 1.04 5.9±0.14ms ? ?/sec 1.00 5.7±0.09ms ? ?/sec
arrow_reader_clickbench/async/Q41 1.04 5.1±0.07ms ? ?/sec 1.00 4.9±0.07ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.02 3.6±0.03ms ? ?/sec 1.00 3.5±0.03ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q1 1.01 1073.2±7.90µs ? ?/sec 1.00 1066.5±1.99µs ? ?/sec
arrow_reader_clickbench/async_object_store/Q10 1.00 6.6±0.09ms ? ?/sec 1.00 6.6±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q11 1.00 7.6±0.20ms ? ?/sec 1.01 7.7±0.15ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q12 1.02 14.6±0.23ms ? ?/sec 1.00 14.3±0.07ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q13 1.00 16.9±0.08ms ? ?/sec 1.02 17.2±0.35ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q14 1.00 16.1±0.30ms ? ?/sec 1.00 16.0±0.24ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q19 1.01 3.0±0.03ms ? ?/sec 1.00 2.9±0.05ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q20 1.02 72.1±0.52ms ? ?/sec 1.00 71.0±0.40ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q21 1.02 81.1±0.77ms ? ?/sec 1.00 79.6±0.37ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q22 1.06 101.9±3.54ms ? ?/sec 1.00 96.6±0.52ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q23 1.00 224.2±2.51ms ? ?/sec 1.02 229.5±1.93ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q24 1.03 19.8±0.26ms ? ?/sec 1.00 19.3±0.23ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q27 1.02 57.5±0.87ms ? ?/sec 1.00 56.6±0.29ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q28 1.01 57.3±0.52ms ? ?/sec 1.00 56.9±0.63ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q30 1.02 18.6±0.36ms ? ?/sec 1.00 18.3±0.06ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q36 1.02 14.8±0.48ms ? ?/sec 1.00 14.6±0.37ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q37 1.04 5.4±0.03ms ? ?/sec 1.00 5.2±0.03ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q38 1.07 13.4±0.41ms ? ?/sec 1.00 12.5±0.17ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q39 1.00 23.4±0.54ms ? ?/sec 1.00 23.5±0.34ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q40 1.02 5.6±0.04ms ? ?/sec 1.00 5.5±0.03ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q41 1.08 5.0±0.07ms ? ?/sec 1.00 4.7±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q42 1.02 3.5±0.02ms ? ?/sec 1.00 3.4±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q1 1.00 874.7±2.79µs ? ?/sec 1.01 881.7±2.48µs ? ?/sec
arrow_reader_clickbench/sync/Q10 1.03 5.3±0.03ms ? ?/sec 1.00 5.1±0.06ms ? ?/sec
arrow_reader_clickbench/sync/Q11 1.01 6.2±0.13ms ? ?/sec 1.00 6.2±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q12 1.05 22.9±0.95ms ? ?/sec 1.00 21.8±0.21ms ? ?/sec
arrow_reader_clickbench/sync/Q13 1.00 29.2±0.83ms ? ?/sec 1.02 29.6±0.42ms ? ?/sec
arrow_reader_clickbench/sync/Q14 1.00 23.6±0.21ms ? ?/sec 1.16 27.2±0.50ms ? ?/sec
arrow_reader_clickbench/sync/Q19 1.04 2.8±0.03ms ? ?/sec 1.00 2.7±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q20 1.01 123.5±1.98ms ? ?/sec 1.00 121.8±0.82ms ? ?/sec
arrow_reader_clickbench/sync/Q21 1.02 98.7±0.52ms ? ?/sec 1.00 96.6±0.32ms ? ?/sec
arrow_reader_clickbench/sync/Q22 1.02 145.7±0.90ms ? ?/sec 1.00 143.5±1.23ms ? ?/sec
arrow_reader_clickbench/sync/Q23 1.06 283.5±13.77ms ? ?/sec 1.00 267.7±7.75ms ? ?/sec
arrow_reader_clickbench/sync/Q24 1.01 27.6±0.34ms ? ?/sec 1.00 27.4±0.25ms ? ?/sec
arrow_reader_clickbench/sync/Q27 1.03 110.1±0.59ms ? ?/sec 1.00 107.1±0.51ms ? ?/sec
arrow_reader_clickbench/sync/Q28 1.02 106.5±0.23ms ? ?/sec 1.00 104.5±0.20ms ? ?/sec
arrow_reader_clickbench/sync/Q30 1.00 19.1±0.20ms ? ?/sec 1.00 19.1±0.18ms ? ?/sec
arrow_reader_clickbench/sync/Q36 1.00 22.6±0.33ms ? ?/sec 1.00 22.6±0.26ms ? ?/sec
arrow_reader_clickbench/sync/Q37 1.01 7.0±0.05ms ? ?/sec 1.00 7.0±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q38 1.02 11.7±0.17ms ? ?/sec 1.00 11.4±0.17ms ? ?/sec
arrow_reader_clickbench/sync/Q39 1.03 21.4±0.13ms ? ?/sec 1.00 20.9±0.08ms ? ?/sec
arrow_reader_clickbench/sync/Q40 1.01 5.4±0.13ms ? ?/sec 1.00 5.4±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q41 1.00 5.8±0.06ms ? ?/sec 1.00 5.8±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q42 1.00 4.4±0.03ms ? ?/sec 1.00 4.4±0.02ms ? ?/sec
```
</p>
</details>
<details><summary>Resource Usage</summary>
**base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 786.5s |
| Peak memory | 3.1 GiB |
| Avg memory | 2.9 GiB |
| CPU user | 707.6s |
| CPU sys | 78.7s |
| Disk read | 0 B |
| Disk write | 846.0 MiB |
**branch**
| Metric | Value |
|--------|-------|
| Wall time | 771.4s |
| Peak memory | 3.2 GiB |
| Avg memory | 3.1 GiB |
| CPU user | 704.4s |
| CPU sys | 66.8s |
| Disk read | 0 B |
| Disk write | 171.3 MiB |
</details>
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096287515
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "alamb-ghbot (via GitHub)" <gi...@apache.org>.
alamb-ghbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3915849322
🤖: Benchmark completed
<details><summary>Details</summary>
<p>
```
group main selectivity_threshold
----- ---- ---------------------
arrow_reader_clickbench/async/Q1 1.00 2.3±0.03ms ? ?/sec 1.01 2.3±0.04ms ? ?/sec
arrow_reader_clickbench/async/Q10 1.00 10.4±0.18ms ? ?/sec 1.02 10.6±0.18ms ? ?/sec
arrow_reader_clickbench/async/Q11 1.01 12.2±0.45ms ? ?/sec 1.00 12.1±0.20ms ? ?/sec
arrow_reader_clickbench/async/Q12 1.00 21.5±0.14ms ? ?/sec 1.04 22.4±0.43ms ? ?/sec
arrow_reader_clickbench/async/Q13 1.00 26.8±0.47ms ? ?/sec 1.03 27.5±0.45ms ? ?/sec
arrow_reader_clickbench/async/Q14 1.00 23.8±0.44ms ? ?/sec 1.04 24.7±0.29ms ? ?/sec
arrow_reader_clickbench/async/Q19 1.00 5.3±0.16ms ? ?/sec 1.00 5.3±0.10ms ? ?/sec
arrow_reader_clickbench/async/Q20 1.00 111.7±0.99ms ? ?/sec 1.02 113.5±1.08ms ? ?/sec
arrow_reader_clickbench/async/Q21 1.00 124.3±0.55ms ? ?/sec 1.01 125.8±0.64ms ? ?/sec
arrow_reader_clickbench/async/Q22 1.00 180.4±7.16ms ? ?/sec 1.05 189.6±5.89ms ? ?/sec
arrow_reader_clickbench/async/Q23 1.00 399.7±2.20ms ? ?/sec 1.02 405.8±4.17ms ? ?/sec
arrow_reader_clickbench/async/Q24 1.00 29.7±0.22ms ? ?/sec 1.03 30.4±0.18ms ? ?/sec
arrow_reader_clickbench/async/Q27 1.00 97.5±0.66ms ? ?/sec 1.01 98.5±0.33ms ? ?/sec
arrow_reader_clickbench/async/Q28 1.00 95.3±0.92ms ? ?/sec 1.01 96.6±0.88ms ? ?/sec
arrow_reader_clickbench/async/Q30 1.00 26.2±0.60ms ? ?/sec 1.04 27.1±0.20ms ? ?/sec
arrow_reader_clickbench/async/Q36 1.08 28.6±0.35ms ? ?/sec 1.00 26.6±0.34ms ? ?/sec
arrow_reader_clickbench/async/Q37 1.28 9.5±0.11ms ? ?/sec 1.00 7.4±0.17ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.00 24.7±0.55ms ? ?/sec 1.00 24.7±0.25ms ? ?/sec
arrow_reader_clickbench/async/Q39 1.00 42.3±0.28ms ? ?/sec 1.02 43.2±0.46ms ? ?/sec
arrow_reader_clickbench/async/Q40 1.37 13.2±0.17ms ? ?/sec 1.00 9.7±0.13ms ? ?/sec
arrow_reader_clickbench/async/Q41 1.31 10.3±0.14ms ? ?/sec 1.00 7.9±0.11ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.06 5.8±0.05ms ? ?/sec 1.00 5.4±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q1 1.00 2.3±0.05ms ? ?/sec 1.00 2.3±0.03ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q10 1.00 10.0±0.15ms ? ?/sec 1.01 10.1±0.19ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q11 1.00 11.5±0.12ms ? ?/sec 1.09 12.5±1.24ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q12 1.00 21.0±0.12ms ? ?/sec 1.03 21.7±0.30ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q13 1.00 25.6±0.19ms ? ?/sec 1.05 26.8±0.40ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q14 1.00 23.4±0.20ms ? ?/sec 1.05 24.5±0.46ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q19 1.00 4.9±0.02ms ? ?/sec 1.04 5.1±0.11ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q20 1.00 107.4±0.40ms ? ?/sec 1.02 110.0±0.62ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q21 1.00 119.9±1.47ms ? ?/sec 1.03 123.3±0.75ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q22 1.00 147.7±1.00ms ? ?/sec 1.05 155.5±4.89ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q23 1.02 365.0±3.70ms ? ?/sec 1.00 356.3±13.16ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q24 1.04 30.5±0.83ms ? ?/sec 1.00 29.4±0.41ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q27 1.00 95.1±1.25ms ? ?/sec 1.00 94.9±0.54ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q28 1.00 92.7±1.24ms ? ?/sec 1.00 92.9±1.48ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q30 1.00 26.9±0.31ms ? ?/sec 1.00 26.8±0.35ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q36 1.11 27.5±0.52ms ? ?/sec 1.00 24.7±0.39ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q37 1.32 9.5±0.11ms ? ?/sec 1.00 7.2±0.07ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q38 1.04 23.0±0.28ms ? ?/sec 1.00 22.1±0.33ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q39 1.00 38.9±0.64ms ? ?/sec 1.02 39.7±0.92ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q40 1.38 13.3±0.26ms ? ?/sec 1.00 9.6±0.37ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q41 1.32 10.1±0.24ms ? ?/sec 1.00 7.7±0.21ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q42 1.08 5.6±0.09ms ? ?/sec 1.00 5.2±0.11ms ? ?/sec
arrow_reader_clickbench/sync/Q1 1.00 1994.7±19.04µs ? ?/sec 1.00 1993.9±8.82µs ? ?/sec
arrow_reader_clickbench/sync/Q10 1.01 7.6±0.12ms ? ?/sec 1.00 7.5±0.04ms ? ?/sec
arrow_reader_clickbench/sync/Q11 1.00 8.8±0.09ms ? ?/sec 1.00 8.8±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q12 1.00 29.1±1.34ms ? ?/sec 1.00 29.2±1.08ms ? ?/sec
arrow_reader_clickbench/sync/Q13 1.00 33.2±0.73ms ? ?/sec 1.20 39.8±1.03ms ? ?/sec
arrow_reader_clickbench/sync/Q14 1.00 30.1±0.26ms ? ?/sec 1.03 30.9±0.14ms ? ?/sec
arrow_reader_clickbench/sync/Q19 1.01 4.2±0.02ms ? ?/sec 1.00 4.1±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q20 1.00 172.2±1.42ms ? ?/sec 1.02 176.1±1.60ms ? ?/sec
arrow_reader_clickbench/sync/Q21 1.00 128.6±0.80ms ? ?/sec 1.02 131.1±1.22ms ? ?/sec
arrow_reader_clickbench/sync/Q22 1.00 205.2±1.96ms ? ?/sec 1.01 206.6±1.81ms ? ?/sec
arrow_reader_clickbench/sync/Q23 1.00 418.9±4.91ms ? ?/sec 1.02 425.5±9.73ms ? ?/sec
arrow_reader_clickbench/sync/Q24 1.00 38.9±1.12ms ? ?/sec 1.02 39.9±0.45ms ? ?/sec
arrow_reader_clickbench/sync/Q27 1.00 148.7±0.86ms ? ?/sec 1.03 152.7±1.50ms ? ?/sec
arrow_reader_clickbench/sync/Q28 1.00 143.1±1.25ms ? ?/sec 1.02 146.6±1.71ms ? ?/sec
arrow_reader_clickbench/sync/Q30 1.00 27.0±0.39ms ? ?/sec 1.01 27.3±0.37ms ? ?/sec
arrow_reader_clickbench/sync/Q36 1.07 32.6±0.77ms ? ?/sec 1.00 30.6±0.23ms ? ?/sec
arrow_reader_clickbench/sync/Q37 1.22 10.4±0.06ms ? ?/sec 1.00 8.5±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q38 1.00 17.8±0.26ms ? ?/sec 1.00 17.8±0.16ms ? ?/sec
arrow_reader_clickbench/sync/Q39 1.00 30.1±0.26ms ? ?/sec 1.02 30.8±0.32ms ? ?/sec
arrow_reader_clickbench/sync/Q40 1.10 9.0±0.08ms ? ?/sec 1.00 8.2±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q41 1.33 9.1±0.05ms ? ?/sec 1.00 6.9±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q42 1.04 6.7±0.04ms ? ?/sec 1.00 6.4±0.13ms ? ?/sec
```
</p>
</details>
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "adriangb (via GitHub)" <gi...@apache.org>.
adriangb commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3914657180
I won't schedule any more benchmarks from that branch. Interesting about the mem usage.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "sdf-jkl (via GitHub)" <gi...@apache.org>.
sdf-jkl commented on code in PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#discussion_r3005452406
##########
parquet/src/arrow/arrow_reader/mod.rs:
##########
@@ -430,6 +434,32 @@ impl<T> ArrowReaderBuilder<T> {
..self
}
}
+
+ /// Set a scatter threshold for filter deferral.
+ ///
+ /// The threshold is the maximum allowed **selector density**
+ /// (`selector_count / row_count`). If applying a predicate would
+ /// produce a density above this value, its result is deferred.
+ /// For example, `0.25` allows at most 25 selectors per 100 rows.
+ ///
+ /// A high selector density means many small skip/read transitions,
Review Comment:
Yes, but low selector density doesn't guarantee that skip/read transitions are large either.
##########
parquet/src/arrow/arrow_reader/mod.rs:
##########
@@ -430,6 +434,32 @@ impl<T> ArrowReaderBuilder<T> {
..self
}
}
+
+ /// Set a scatter threshold for filter deferral.
Review Comment:
I think it's great, but this heuristic doesn't know anything about the distribution of data. Low selector density doesn't guarantee that skip/read transitions are large.
For example:
```
total_rows = 10_000
selector_count = 100
density = 100 / 10_000 = 0.01 (at threshold)
RowSelection (100 selectors total):
1) select 4951 rows
2) skip 1
3) select 1
4) skip 1
...
99) skip 1
100) select 4951 rows
```
Which wouldn't be good:
```
[ SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS ][XSXSXSXSXS...XS][ SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS ]
~4951 selected 98 rows ~4951 selected
```
One way to catch this is by checking selectivity. `49 / 10_000 = 0.0049` (Very low)
----
I propose keeping both `selector_density` and `selectivity_threshold` you had before.
This matrix shows the possible cases:
| Density (`selectors/rows`) | Selectivity (`skipped/rows`) | Typical Pattern | Pushdown / Deferral Hint |
|---|---|---|---|
| Low | Low | Small contiguous skipped island | Would think is good without checking the selectivity |
| Low | High | Large contiguous skips | Great!|
| High | Low | Many tiny transitions, little skipped overall | Worst case |
| High | High | Many tiny transitions but lots skipped | Check if it increases the `selector_count` |
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "sdf-jkl (via GitHub)" <gi...@apache.org>.
sdf-jkl commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4159269629
I also now have a very different idea.
Basically, why deal with averages around densities and selections, when we have the real numbers?
For selection materialization we loop through the selectors and count non-zero ones:
https://github.com/apache/arrow-rs/blob/aa9432c8833f5701085e8b933b30560d21df9f80/parquet/src/arrow/arrow_reader/read_plan.rs#L113-L123
Why don't we count rows covered by long/short selectors instead?
```rust
// total_rows: total rows covered by selection
// effective_count: non-empty selector runs
// short_*_rows: rows in runs <= short_threshold
// long_*_rows: rows in runs >= long_threshold
let (
total_rows,
effective_count,
short_select_rows,
short_skip_rows,
long_select_rows,
long_skip_rows,
) = selection.iter().fold(
(0usize, 0usize, 0usize, 0usize, 0usize, 0usize),
|(rows, cnt, ss, sk, ls, lk), s| {
if s.row_count == 0 {
return (rows, cnt, ss, sk, ls, lk);
}
let rows = rows + s.row_count;
let cnt = cnt + 1;
let is_short = s.row_count <= short_threshold;
let is_long = s.row_count >= long_threshold;
match (s.skip, is_short, is_long) {
(true, true, _) => (rows, cnt, ss, sk + s.row_count, ls, lk),
(true, _, true) => (rows, cnt, ss, sk, ls, lk + s.row_count),
(false, true, _) => (rows, cnt, ss + s.row_count, sk, ls, lk),
(false, _, true) => (rows, cnt, ss, sk, ls + s.row_count, lk),
_ => (rows, cnt, ss, sk, ls, lk), // middle bin
}
},
);
```
Where `long/short_threshold` would be based on the page size or something else?
With this statistics/histogram we'd be able to make a more data-driven decision on keeping or deferring the selection.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] [PoC] Filter pushdown selectivity threshold [arrow-rs]
Posted by "alamb-ghbot (via GitHub)" <gi...@apache.org>.
alamb-ghbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3902487470
🤖: Benchmark completed
<details><summary>Details</summary>
<p>
```
group main selectivity_threshold
----- ---- ---------------------
arrow_reader_clickbench/async/Q1 1.00 2.3±0.03ms ? ?/sec 1.00 2.3±0.02ms ? ?/sec
arrow_reader_clickbench/async/Q10 1.01 11.6±0.40ms ? ?/sec 1.00 11.5±0.40ms ? ?/sec
arrow_reader_clickbench/async/Q11 1.00 13.0±0.27ms ? ?/sec 1.03 13.4±0.46ms ? ?/sec
arrow_reader_clickbench/async/Q12 1.00 23.1±0.38ms ? ?/sec 1.01 23.5±0.35ms ? ?/sec
arrow_reader_clickbench/async/Q13 1.00 28.5±0.28ms ? ?/sec 1.01 28.9±0.41ms ? ?/sec
arrow_reader_clickbench/async/Q14 1.00 25.5±0.19ms ? ?/sec 1.02 26.0±0.45ms ? ?/sec
arrow_reader_clickbench/async/Q19 1.00 5.6±0.10ms ? ?/sec 1.04 5.8±0.33ms ? ?/sec
arrow_reader_clickbench/async/Q20 1.12 128.4±11.83ms ? ?/sec 1.00 114.8±0.57ms ? ?/sec
arrow_reader_clickbench/async/Q21 1.30 166.3±1.04ms ? ?/sec 1.00 128.1±1.42ms ? ?/sec
arrow_reader_clickbench/async/Q22 1.31 241.9±1.35ms ? ?/sec 1.00 185.3±4.95ms ? ?/sec
arrow_reader_clickbench/async/Q23 1.00 412.2±2.41ms ? ?/sec 1.00 412.5±5.81ms ? ?/sec
arrow_reader_clickbench/async/Q24 1.00 32.1±0.70ms ? ?/sec 1.00 32.1±0.31ms ? ?/sec
arrow_reader_clickbench/async/Q27 1.00 100.9±0.68ms ? ?/sec 1.00 100.6±0.66ms ? ?/sec
arrow_reader_clickbench/async/Q28 1.00 98.7±1.00ms ? ?/sec 1.00 98.3±0.63ms ? ?/sec
arrow_reader_clickbench/async/Q30 1.00 28.1±0.32ms ? ?/sec 1.00 28.1±0.28ms ? ?/sec
arrow_reader_clickbench/async/Q36 4.26 118.8±0.61ms ? ?/sec 1.00 27.9±0.51ms ? ?/sec
arrow_reader_clickbench/async/Q37 12.22 93.9±0.68ms ? ?/sec 1.00 7.7±0.13ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.40 35.7±0.81ms ? ?/sec 1.00 25.5±0.24ms ? ?/sec
arrow_reader_clickbench/async/Q39 1.04 46.2±0.32ms ? ?/sec 1.00 44.5±0.47ms ? ?/sec
arrow_reader_clickbench/async/Q40 3.95 40.7±0.51ms ? ?/sec 1.00 10.3±0.25ms ? ?/sec
arrow_reader_clickbench/async/Q41 3.48 29.8±0.44ms ? ?/sec 1.00 8.6±0.24ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.97 11.0±0.07ms ? ?/sec 1.00 5.6±0.10ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q1 1.00 2.2±0.04ms ? ?/sec 1.01 2.3±0.05ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q10 1.00 11.2±0.30ms ? ?/sec 1.00 11.2±0.31ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q11 1.01 12.7±0.37ms ? ?/sec 1.00 12.6±0.41ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q12 1.00 22.5±0.20ms ? ?/sec 1.00 22.6±0.30ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q13 1.00 27.6±0.31ms ? ?/sec 1.00 27.7±0.28ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q14 1.00 25.0±0.27ms ? ?/sec 1.00 25.1±0.25ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q19 1.00 5.2±0.09ms ? ?/sec 1.01 5.2±0.11ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q20 1.00 110.9±1.31ms ? ?/sec 1.00 110.7±1.00ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q21 1.03 126.8±0.89ms ? ?/sec 1.00 123.3±1.12ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q22 1.52 233.5±1.32ms ? ?/sec 1.00 153.2±0.99ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q23 1.04 361.3±24.92ms ? ?/sec 1.00 347.2±1.65ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q24 1.00 30.2±0.29ms ? ?/sec 1.02 30.8±0.50ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q27 1.00 96.1±0.81ms ? ?/sec 1.00 95.8±0.62ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q28 1.00 94.2±1.01ms ? ?/sec 1.00 94.3±0.64ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q30 1.00 26.9±0.25ms ? ?/sec 1.01 27.1±0.29ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q36 4.53 113.6±0.58ms ? ?/sec 1.00 25.1±0.26ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q37 12.06 90.0±0.61ms ? ?/sec 1.00 7.5±0.10ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q38 1.43 32.6±0.67ms ? ?/sec 1.00 22.8±0.28ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q39 1.05 41.5±0.31ms ? ?/sec 1.00 39.7±0.37ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q40 4.07 39.0±0.68ms ? ?/sec 1.00 9.6±0.17ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q41 3.55 28.2±0.30ms ? ?/sec 1.00 7.9±0.15ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q42 1.98 10.5±0.12ms ? ?/sec 1.00 5.3±0.04ms ? ?/sec
arrow_reader_clickbench/sync/Q1 1.00 1983.5±13.50µs ? ?/sec 1.01 1993.7±15.84µs ? ?/sec
arrow_reader_clickbench/sync/Q10 1.00 7.5±0.08ms ? ?/sec 1.00 7.6±0.10ms ? ?/sec
arrow_reader_clickbench/sync/Q11 1.00 9.0±0.07ms ? ?/sec 1.00 9.0±0.32ms ? ?/sec
arrow_reader_clickbench/sync/Q12 1.00 29.6±1.36ms ? ?/sec 1.02 30.1±1.19ms ? ?/sec
arrow_reader_clickbench/sync/Q13 1.00 34.3±0.67ms ? ?/sec 1.20 41.3±1.39ms ? ?/sec
arrow_reader_clickbench/sync/Q14 1.23 39.1±1.01ms ? ?/sec 1.00 31.8±0.32ms ? ?/sec
arrow_reader_clickbench/sync/Q19 1.00 4.2±0.04ms ? ?/sec 1.01 4.2±0.08ms ? ?/sec
arrow_reader_clickbench/sync/Q20 1.01 179.3±1.24ms ? ?/sec 1.00 177.6±2.26ms ? ?/sec
arrow_reader_clickbench/sync/Q21 1.73 231.3±5.30ms ? ?/sec 1.00 133.8±1.14ms ? ?/sec
arrow_reader_clickbench/sync/Q22 2.30 481.2±3.88ms ? ?/sec 1.00 208.9±1.42ms ? ?/sec
arrow_reader_clickbench/sync/Q23 1.06 448.8±16.62ms ? ?/sec 1.00 422.6±6.25ms ? ?/sec
arrow_reader_clickbench/sync/Q24 1.02 39.9±0.35ms ? ?/sec 1.00 39.0±0.33ms ? ?/sec
arrow_reader_clickbench/sync/Q27 1.00 154.8±1.28ms ? ?/sec 1.01 156.1±1.15ms ? ?/sec
arrow_reader_clickbench/sync/Q28 1.00 149.5±1.60ms ? ?/sec 1.01 150.4±0.98ms ? ?/sec
arrow_reader_clickbench/sync/Q30 1.00 28.3±0.40ms ? ?/sec 1.00 28.4±0.30ms ? ?/sec
arrow_reader_clickbench/sync/Q36 4.77 153.9±1.60ms ? ?/sec 1.00 32.2±0.49ms ? ?/sec
arrow_reader_clickbench/sync/Q37 9.86 85.9±0.63ms ? ?/sec 1.00 8.7±0.07ms ? ?/sec
arrow_reader_clickbench/sync/Q38 1.53 28.4±0.46ms ? ?/sec 1.00 18.5±0.24ms ? ?/sec
arrow_reader_clickbench/sync/Q39 1.06 34.0±0.45ms ? ?/sec 1.00 32.0±0.32ms ? ?/sec
arrow_reader_clickbench/sync/Q40 3.06 26.6±0.58ms ? ?/sec 1.00 8.7±0.28ms ? ?/sec
arrow_reader_clickbench/sync/Q41 3.92 27.9±0.30ms ? ?/sec 1.00 7.1±0.09ms ? ?/sec
arrow_reader_clickbench/sync/Q42 1.76 11.8±0.10ms ? ?/sec 1.00 6.7±0.23ms ? ?/sec
```
</p>
</details>
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095547902
> Looking at the code here I think this may make sense regardless of the work on the DataFusion side. It's a relatively small self contained change. Even if we move filters from row filters to non row filters (which has implications for IO, etc.) this at the very least saves compute applying pointless fragmented masks (if I understand the change correctly), right?
Yes, I think it's pretty effective, simple, and does something DataFusion can not handle yet (skip applying the mask indeed).
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096077282
🤖 Arrow criterion benchmark running (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096060759)
`Linux bench-c4096060759-476-hl5d8 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux`
Comparing selectivity_threshold (a9bc94fa6c2dd725288847e2f7d0c64e40ac92bc) to d53df60 (merge-base) [diff](https://github.com/apache/arrow-rs/compare/d53df605656d8012eca42e8ddffe165362a1a4cb..a9bc94fa6c2dd725288847e2f7d0c64e40ac92bc)
BENCH_NAME=arrow_reader_clickbench
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench arrow_reader_clickbench
BENCH_FILTER=
Results will be posted here when complete
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] [PoC] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3902519361
Nice this shows some good resulrs @alamb @adriangb (need to merge https://github.com/apache/arrow-rs/pull/9413) so we can run the benchmark again and compare.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3914621343
> @alamb can you get the runner "unstuck" again? :D
Done
What I think is happening is that the runer is being oomkilled:
> [9206835.020555] Out of memory: Killed process 2715082 (dfbench) total-vm:47838332kB, anon-rss:31959228kB, file-rss:2792kB, shmem-rss:0kB, UID:1001 pgtables:63624kB oom_score_adj:0
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on code in PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#discussion_r2966107445
##########
parquet/src/arrow/arrow_reader/mod.rs:
##########
@@ -430,6 +434,32 @@ impl<T> ArrowReaderBuilder<T> {
..self
}
}
+
+ /// Set a scatter threshold for filter deferral.
Review Comment:
Yeah - I also think this is quite a bit better - I think it's OK if a filter "only" selects 50% of the rows, if it is nicely packed that will probably be able to skip almost 50% of the pages and lead to efficient IO and decoding.
But (especially when combining multiple filters), chances are the filter gets too fragmented, a 50% packed one is better than a 25% selective one that just does select 1 / skip 1 / select 1 ...
I think you could go even go further and check which combination of filtera are giving the best of both worlds (selectivity and fragmentation)
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096433401
🤖 Arrow criterion benchmark completed (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096289927)
<details><summary>Details</summary>
<p>
```
group main selectivity_threshold
----- ---- ---------------------
arrow_reader_clickbench/async/Q1 1.00 1090.2±4.95µs ? ?/sec 1.00 1094.3±5.40µs ? ?/sec
arrow_reader_clickbench/async/Q10 1.01 6.6±0.04ms ? ?/sec 1.00 6.6±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q11 1.01 7.7±0.04ms ? ?/sec 1.00 7.6±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q12 1.02 14.5±0.11ms ? ?/sec 1.00 14.3±0.09ms ? ?/sec
arrow_reader_clickbench/async/Q13 1.02 17.1±0.12ms ? ?/sec 1.00 16.8±0.17ms ? ?/sec
arrow_reader_clickbench/async/Q14 1.03 16.1±0.15ms ? ?/sec 1.00 15.7±0.10ms ? ?/sec
arrow_reader_clickbench/async/Q19 1.00 3.1±0.02ms ? ?/sec 1.00 3.0±0.02ms ? ?/sec
arrow_reader_clickbench/async/Q20 1.09 78.7±0.30ms ? ?/sec 1.00 71.9±0.31ms ? ?/sec
arrow_reader_clickbench/async/Q21 1.21 97.5±0.83ms ? ?/sec 1.00 80.4±0.28ms ? ?/sec
arrow_reader_clickbench/async/Q22 1.16 132.1±5.00ms ? ?/sec 1.00 114.0±2.41ms ? ?/sec
arrow_reader_clickbench/async/Q23 1.01 244.4±7.51ms ? ?/sec 1.00 243.1±4.05ms ? ?/sec
arrow_reader_clickbench/async/Q24 1.00 19.4±0.15ms ? ?/sec 1.00 19.4±0.12ms ? ?/sec
arrow_reader_clickbench/async/Q27 1.00 57.6±0.32ms ? ?/sec 1.00 57.9±0.45ms ? ?/sec
arrow_reader_clickbench/async/Q28 1.02 58.1±0.32ms ? ?/sec 1.00 57.0±0.28ms ? ?/sec
arrow_reader_clickbench/async/Q30 1.01 18.7±0.10ms ? ?/sec 1.00 18.5±0.07ms ? ?/sec
arrow_reader_clickbench/async/Q36 1.08 15.2±0.30ms ? ?/sec 1.00 14.1±0.27ms ? ?/sec
arrow_reader_clickbench/async/Q37 1.18 5.4±0.01ms ? ?/sec 1.00 4.6±0.01ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.13 14.4±0.88ms ? ?/sec 1.00 12.8±0.19ms ? ?/sec
arrow_reader_clickbench/async/Q39 1.05 25.9±0.90ms ? ?/sec 1.00 24.7±0.72ms ? ?/sec
arrow_reader_clickbench/async/Q40 1.13 5.7±0.04ms ? ?/sec 1.00 5.1±0.03ms ? ?/sec
arrow_reader_clickbench/async/Q41 1.11 5.0±0.03ms ? ?/sec 1.00 4.5±0.02ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.10 3.6±0.02ms ? ?/sec 1.00 3.3±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q1 1.00 1068.8±8.50µs ? ?/sec 1.00 1071.6±3.39µs ? ?/sec
arrow_reader_clickbench/async_object_store/Q10 1.08 7.1±0.18ms ? ?/sec 1.00 6.6±0.03ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q11 1.05 7.9±0.13ms ? ?/sec 1.00 7.5±0.05ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q12 1.07 15.1±0.13ms ? ?/sec 1.00 14.1±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q13 1.06 17.8±0.22ms ? ?/sec 1.00 16.9±0.21ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q14 1.04 16.4±0.14ms ? ?/sec 1.00 15.8±0.12ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q19 1.00 2.9±0.02ms ? ?/sec 1.00 2.9±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q20 1.00 71.0±0.49ms ? ?/sec 1.00 71.2±0.36ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q21 1.00 80.2±0.59ms ? ?/sec 1.00 80.0±0.37ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q22 1.00 97.5±1.21ms ? ?/sec 1.02 99.7±3.72ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q23 1.05 232.8±1.81ms ? ?/sec 1.00 221.0±0.58ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q24 1.00 19.2±0.14ms ? ?/sec 1.00 19.2±0.16ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q27 1.00 56.2±0.38ms ? ?/sec 1.01 56.9±0.36ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q28 1.02 57.5±0.45ms ? ?/sec 1.00 56.5±0.31ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q30 1.02 18.4±0.13ms ? ?/sec 1.00 18.1±0.08ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q36 1.06 14.4±0.24ms ? ?/sec 1.00 13.6±0.20ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q37 1.18 5.3±0.01ms ? ?/sec 1.00 4.5±0.01ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q38 1.05 12.7±0.21ms ? ?/sec 1.00 12.0±0.19ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q39 1.00 23.4±0.45ms ? ?/sec 1.00 23.3±0.44ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q40 1.11 5.5±0.02ms ? ?/sec 1.00 4.9±0.03ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q41 1.10 4.8±0.02ms ? ?/sec 1.00 4.4±0.02ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q42 1.08 3.4±0.01ms ? ?/sec 1.00 3.2±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q1 1.01 876.8±3.83µs ? ?/sec 1.00 868.7±2.14µs ? ?/sec
arrow_reader_clickbench/sync/Q10 1.01 5.1±0.02ms ? ?/sec 1.00 5.1±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q11 1.00 6.1±0.02ms ? ?/sec 1.00 6.1±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q12 1.03 22.1±0.60ms ? ?/sec 1.00 21.5±0.14ms ? ?/sec
arrow_reader_clickbench/sync/Q13 1.00 28.3±0.80ms ? ?/sec 1.04 29.4±0.29ms ? ?/sec
arrow_reader_clickbench/sync/Q14 1.00 23.3±0.09ms ? ?/sec 1.15 26.9±0.43ms ? ?/sec
arrow_reader_clickbench/sync/Q19 1.04 2.7±0.03ms ? ?/sec 1.00 2.6±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q20 1.01 123.2±0.38ms ? ?/sec 1.00 122.3±0.73ms ? ?/sec
arrow_reader_clickbench/sync/Q21 1.01 98.1±0.25ms ? ?/sec 1.00 96.8±0.30ms ? ?/sec
arrow_reader_clickbench/sync/Q22 1.00 142.8±0.36ms ? ?/sec 1.01 144.2±0.72ms ? ?/sec
arrow_reader_clickbench/sync/Q23 1.05 282.5±16.04ms ? ?/sec 1.00 268.5±7.96ms ? ?/sec
arrow_reader_clickbench/sync/Q24 1.02 27.2±0.13ms ? ?/sec 1.00 26.6±0.12ms ? ?/sec
arrow_reader_clickbench/sync/Q27 1.01 107.2±0.99ms ? ?/sec 1.00 105.9±0.40ms ? ?/sec
arrow_reader_clickbench/sync/Q28 1.04 106.8±0.17ms ? ?/sec 1.00 102.9±0.35ms ? ?/sec
arrow_reader_clickbench/sync/Q30 1.03 19.0±0.25ms ? ?/sec 1.00 18.6±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q36 1.09 22.5±0.09ms ? ?/sec 1.00 20.7±0.06ms ? ?/sec
arrow_reader_clickbench/sync/Q37 1.18 6.9±0.01ms ? ?/sec 1.00 5.9±0.03ms ? ?/sec
arrow_reader_clickbench/sync/Q38 1.01 11.2±0.05ms ? ?/sec 1.00 11.2±0.06ms ? ?/sec
arrow_reader_clickbench/sync/Q39 1.03 20.9±0.26ms ? ?/sec 1.00 20.4±0.06ms ? ?/sec
arrow_reader_clickbench/sync/Q40 1.00 5.2±0.03ms ? ?/sec 1.09 5.7±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q41 1.11 5.6±0.03ms ? ?/sec 1.00 5.1±0.01ms ? ?/sec
arrow_reader_clickbench/sync/Q42 1.00 4.4±0.02ms ? ?/sec 1.01 4.5±0.01ms ? ?/sec
```
</p>
</details>
<details><summary>Resource Usage</summary>
**base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 782.8s |
| Peak memory | 3.1 GiB |
| Avg memory | 2.9 GiB |
| CPU user | 707.0s |
| CPU sys | 75.5s |
| Disk read | 0 B |
| Disk write | 2.1 GiB |
**branch**
| Metric | Value |
|--------|-------|
| Wall time | 763.2s |
| Peak memory | 3.2 GiB |
| Avg memory | 3.1 GiB |
| CPU user | 697.7s |
| CPU sys | 65.5s |
| Disk read | 0 B |
| Disk write | 171.4 MiB |
</details>
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096060759
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096289927
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] [PoC] Selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3902397936
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4785346799
Thank you for your contribution. Unfortunately, this pull request is stale because it has been open 60 days with no activity. Please remove the stale label or comment or this will be closed in 7 days.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3915640801
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4093064693
Actually I think we should just have / try this.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095543573
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096433508
🤖 Arrow criterion benchmark completed (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096287515)
<details><summary>Details</summary>
<p>
```
group main selectivity_threshold
----- ---- ---------------------
arrow_reader_clickbench/async/Q1 1.01 1101.3±7.70µs ? ?/sec 1.00 1090.2±9.59µs ? ?/sec
arrow_reader_clickbench/async/Q10 1.01 6.8±0.26ms ? ?/sec 1.00 6.8±0.24ms ? ?/sec
arrow_reader_clickbench/async/Q11 1.00 7.9±0.28ms ? ?/sec 1.00 7.9±0.25ms ? ?/sec
arrow_reader_clickbench/async/Q12 1.01 15.1±0.24ms ? ?/sec 1.00 14.9±0.05ms ? ?/sec
arrow_reader_clickbench/async/Q13 1.03 17.6±0.32ms ? ?/sec 1.00 17.2±0.38ms ? ?/sec
arrow_reader_clickbench/async/Q14 1.01 16.3±0.39ms ? ?/sec 1.00 16.1±0.34ms ? ?/sec
arrow_reader_clickbench/async/Q19 1.00 3.1±0.06ms ? ?/sec 1.00 3.1±0.06ms ? ?/sec
arrow_reader_clickbench/async/Q20 1.24 89.7±1.27ms ? ?/sec 1.00 72.1±0.54ms ? ?/sec
arrow_reader_clickbench/async/Q21 1.29 104.9±1.85ms ? ?/sec 1.00 81.1±0.40ms ? ?/sec
arrow_reader_clickbench/async/Q22 1.19 134.5±4.84ms ? ?/sec 1.00 112.7±2.82ms ? ?/sec
arrow_reader_clickbench/async/Q23 1.02 245.1±2.54ms ? ?/sec 1.00 240.6±2.28ms ? ?/sec
arrow_reader_clickbench/async/Q24 1.02 20.0±0.39ms ? ?/sec 1.00 19.6±0.41ms ? ?/sec
arrow_reader_clickbench/async/Q27 1.01 59.0±0.26ms ? ?/sec 1.00 58.3±0.33ms ? ?/sec
arrow_reader_clickbench/async/Q28 1.02 58.9±0.77ms ? ?/sec 1.00 57.7±0.63ms ? ?/sec
arrow_reader_clickbench/async/Q30 1.01 18.9±0.26ms ? ?/sec 1.00 18.7±0.20ms ? ?/sec
arrow_reader_clickbench/async/Q36 1.06 14.8±0.44ms ? ?/sec 1.00 14.0±0.47ms ? ?/sec
arrow_reader_clickbench/async/Q37 1.18 5.5±0.10ms ? ?/sec 1.00 4.6±0.04ms ? ?/sec
arrow_reader_clickbench/async/Q38 1.03 13.3±0.32ms ? ?/sec 1.00 12.9±0.35ms ? ?/sec
arrow_reader_clickbench/async/Q39 1.00 24.1±0.18ms ? ?/sec 1.01 24.4±0.50ms ? ?/sec
arrow_reader_clickbench/async/Q40 1.10 5.7±0.11ms ? ?/sec 1.00 5.2±0.10ms ? ?/sec
arrow_reader_clickbench/async/Q41 1.09 5.0±0.07ms ? ?/sec 1.00 4.6±0.07ms ? ?/sec
arrow_reader_clickbench/async/Q42 1.09 3.6±0.05ms ? ?/sec 1.00 3.3±0.04ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q1 1.00 1074.5±6.25µs ? ?/sec 1.00 1070.0±3.72µs ? ?/sec
arrow_reader_clickbench/async_object_store/Q10 1.00 6.7±0.20ms ? ?/sec 1.00 6.7±0.24ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q11 1.03 8.0±0.14ms ? ?/sec 1.00 7.8±0.25ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q12 1.00 14.7±0.35ms ? ?/sec 1.00 14.7±0.19ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q13 1.02 17.3±0.46ms ? ?/sec 1.00 16.9±0.45ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q14 1.03 16.4±0.43ms ? ?/sec 1.00 15.9±0.37ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q19 1.00 3.0±0.05ms ? ?/sec 1.03 3.1±0.06ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q20 1.01 71.9±0.59ms ? ?/sec 1.00 71.2±0.60ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q21 1.01 81.1±0.64ms ? ?/sec 1.00 80.1±0.50ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q22 1.01 97.8±1.02ms ? ?/sec 1.00 96.8±0.88ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q23 1.00 221.8±2.38ms ? ?/sec 1.04 229.5±6.69ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q24 1.03 20.1±0.35ms ? ?/sec 1.00 19.5±0.43ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q27 1.00 57.3±0.69ms ? ?/sec 1.00 57.2±0.52ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q28 1.01 57.5±0.98ms ? ?/sec 1.00 57.2±0.53ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q30 1.01 18.5±0.29ms ? ?/sec 1.00 18.3±0.23ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q36 1.08 14.5±0.50ms ? ?/sec 1.00 13.4±0.59ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q37 1.20 5.5±0.09ms ? ?/sec 1.00 4.6±0.05ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q38 1.05 12.9±0.32ms ? ?/sec 1.00 12.3±0.44ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q39 1.00 23.0±0.41ms ? ?/sec 1.03 23.7±0.30ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q40 1.12 5.6±0.12ms ? ?/sec 1.00 5.0±0.15ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q41 1.12 4.9±0.08ms ? ?/sec 1.00 4.4±0.09ms ? ?/sec
arrow_reader_clickbench/async_object_store/Q42 1.09 3.5±0.05ms ? ?/sec 1.00 3.2±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q1 1.02 886.2±16.66µs ? ?/sec 1.00 871.8±2.41µs ? ?/sec
arrow_reader_clickbench/sync/Q10 1.01 5.2±0.06ms ? ?/sec 1.00 5.2±0.05ms ? ?/sec
arrow_reader_clickbench/sync/Q11 1.01 6.2±0.07ms ? ?/sec 1.00 6.1±0.08ms ? ?/sec
arrow_reader_clickbench/sync/Q12 1.02 22.2±0.44ms ? ?/sec 1.00 21.8±0.28ms ? ?/sec
arrow_reader_clickbench/sync/Q13 1.00 24.8±0.43ms ? ?/sec 1.21 30.1±0.82ms ? ?/sec
arrow_reader_clickbench/sync/Q14 1.00 23.6±0.38ms ? ?/sec 1.16 27.3±0.56ms ? ?/sec
arrow_reader_clickbench/sync/Q19 1.00 2.7±0.05ms ? ?/sec 1.00 2.7±0.04ms ? ?/sec
arrow_reader_clickbench/sync/Q20 1.01 123.9±0.74ms ? ?/sec 1.00 122.2±1.15ms ? ?/sec
arrow_reader_clickbench/sync/Q21 1.01 98.1±1.01ms ? ?/sec 1.00 97.4±0.94ms ? ?/sec
arrow_reader_clickbench/sync/Q22 1.00 144.0±2.14ms ? ?/sec 1.01 145.0±1.88ms ? ?/sec
arrow_reader_clickbench/sync/Q23 1.04 280.2±13.85ms ? ?/sec 1.00 270.6±7.87ms ? ?/sec
arrow_reader_clickbench/sync/Q24 1.01 27.5±0.55ms ? ?/sec 1.00 27.2±0.47ms ? ?/sec
arrow_reader_clickbench/sync/Q27 1.01 108.0±1.05ms ? ?/sec 1.00 107.2±0.84ms ? ?/sec
arrow_reader_clickbench/sync/Q28 1.02 106.3±0.87ms ? ?/sec 1.00 104.3±0.51ms ? ?/sec
arrow_reader_clickbench/sync/Q30 1.01 19.4±0.05ms ? ?/sec 1.00 19.2±0.11ms ? ?/sec
arrow_reader_clickbench/sync/Q36 1.05 22.3±0.41ms ? ?/sec 1.00 21.2±0.41ms ? ?/sec
arrow_reader_clickbench/sync/Q37 1.18 7.0±0.09ms ? ?/sec 1.00 5.9±0.07ms ? ?/sec
arrow_reader_clickbench/sync/Q38 1.00 11.4±0.22ms ? ?/sec 1.00 11.5±0.19ms ? ?/sec
arrow_reader_clickbench/sync/Q39 1.00 21.0±0.34ms ? ?/sec 1.01 21.1±0.26ms ? ?/sec
arrow_reader_clickbench/sync/Q40 1.00 5.3±0.09ms ? ?/sec 1.12 5.9±0.02ms ? ?/sec
arrow_reader_clickbench/sync/Q41 1.10 5.7±0.08ms ? ?/sec 1.00 5.2±0.08ms ? ?/sec
arrow_reader_clickbench/sync/Q42 1.00 4.4±0.08ms ? ?/sec 1.03 4.5±0.06ms ? ?/sec
```
</p>
</details>
<details><summary>Resource Usage</summary>
**base (merge-base)**
| Metric | Value |
|--------|-------|
| Wall time | 789.1s |
| Peak memory | 3.1 GiB |
| Avg memory | 2.9 GiB |
| CPU user | 712.6s |
| CPU sys | 76.4s |
| Disk read | 0 B |
| Disk write | 2.0 GiB |
**branch**
| Metric | Value |
|--------|-------|
| Wall time | 767.8s |
| Peak memory | 3.2 GiB |
| Avg memory | 3.1 GiB |
| CPU user | 699.9s |
| CPU sys | 67.9s |
| Disk read | 0 B |
| Disk write | 171.3 MiB |
</details>
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] [PoC] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3904516323
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4093066943
run bencmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095977120
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096300643
🤖 Arrow criterion benchmark running (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096287515)
`Linux bench-c4096287515-477-r5znv 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux`
Comparing selectivity_threshold (9cc2660a005eace5565315e3276194368ced1228) to d53df60 (merge-base) [diff](https://github.com/apache/arrow-rs/compare/d53df605656d8012eca42e8ddffe165362a1a4cb..9cc2660a005eace5565315e3276194368ced1228)
BENCH_NAME=arrow_reader_clickbench
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench arrow_reader_clickbench
BENCH_FILTER=
Results will be posted here when complete
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096302619
🤖 Arrow criterion benchmark running (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4096289927)
`Linux bench-c4096289927-478-9ltb4 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux`
Comparing selectivity_threshold (9cc2660a005eace5565315e3276194368ced1228) to d53df60 (merge-base) [diff](https://github.com/apache/arrow-rs/compare/d53df605656d8012eca42e8ddffe165362a1a4cb..9cc2660a005eace5565315e3276194368ced1228)
BENCH_NAME=arrow_reader_clickbench
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench arrow_reader_clickbench
BENCH_FILTER=
Results will be posted here when complete
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3907223390
run benchmark arrow_reader_clickbench
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] [PoC] Filter pushdown selectivity threshold [arrow-rs]
Posted by "alamb-ghbot (via GitHub)" <gi...@apache.org>.
alamb-ghbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3902440563
🤖 `./gh_compare_arrow.sh` [gh_compare_arrow.sh](https://github.com/alamb/datafusion-benchmarking/blob/main/scripts/gh_compare_arrow.sh) Running
Linux aal-dev 6.14.0-1018-gcp #19~24.04.1-Ubuntu SMP Wed Sep 24 23:23:09 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
Comparing selectivity_threshold (b4275cbf45394f7b8e2eac628b32490dd07f3712) to 39a2b71e55e3fa12ee06defb1d133f828bb383f3 [diff](https://github.com/apache/arrow-rs/compare/39a2b71e55e3fa12ee06defb1d133f828bb383f3..b4275cbf45394f7b8e2eac628b32490dd07f3712)
BENCH_NAME=arrow_reader_clickbench
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench arrow_reader_clickbench
BENCH_FILTER=
BENCH_BRANCH_NAME=selectivity_threshold
Results will be posted here when complete
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095552878
🤖 Arrow criterion benchmark running (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095543573)
`Linux bench-c4095543573-474-jlx7m 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux`
Comparing selectivity_threshold (155a1fc14e6e5b2b4c255c282467c59db1230c1c) to d53df60 (merge-base) [diff](https://github.com/apache/arrow-rs/compare/d53df605656d8012eca42e8ddffe165362a1a4cb..155a1fc14e6e5b2b4c255c282467c59db1230c1c)
BENCH_NAME=arrow_reader_clickbench
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench arrow_reader_clickbench
BENCH_FILTER=
Results will be posted here when complete
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "adriangb (via GitHub)" <gi...@apache.org>.
adriangb commented on code in PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#discussion_r2965818088
##########
parquet/src/arrow/arrow_reader/mod.rs:
##########
@@ -430,6 +434,32 @@ impl<T> ArrowReaderBuilder<T> {
..self
}
}
+
+ /// Set a scatter threshold for filter deferral.
Review Comment:
I like this even more now! This is now dealing with selector density, something that would be hard to do on the datafusion side and is correlated but not equivalent to overall filter selectivity, which I think is what probably ends up mattering more for coarser IO patterns and such given object store range coalescing, pages, etc.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "adriangb (via GitHub)" <gi...@apache.org>.
adriangb commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4093094883
Looking at the code here I think this may make sense regardless of the work on the DataFusion side. It's a relatively small self contained change. Even if we move filters from row filters to non row filters (which has implications for IO, etc.) this at the very least saves compute applying pointless fragmented masks (if I understand the change correctly), right?
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on code in PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#discussion_r2966107445
##########
parquet/src/arrow/arrow_reader/mod.rs:
##########
@@ -430,6 +434,32 @@ impl<T> ArrowReaderBuilder<T> {
..self
}
}
+
+ /// Set a scatter threshold for filter deferral.
Review Comment:
Yeah - I also think this is quite a bit better - I think it's OK if a filter "only" selects 50% of the rows, if it is nicely packed that will probably be able to skip almost 50% of the pages and lead to efficient IO and decoding.
But (especially when combining multiple filters), chances are the filter gets too fragmented, a 50% packed one is better than a 25% selective one that just does select 1 / skip 1 / select 1 and will do the same amount of IO and having horrible decoding performance.
I think you could go even go further and check which combination of filtera are giving the best of both worlds (selectivity and fragmentation)
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Add filter pushdown scatter threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on code in PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#discussion_r2966107445
##########
parquet/src/arrow/arrow_reader/mod.rs:
##########
@@ -430,6 +434,32 @@ impl<T> ArrowReaderBuilder<T> {
..self
}
}
+
+ /// Set a scatter threshold for filter deferral.
Review Comment:
Yeah - I also think this is quite a bit better - I think it's OK if a filter "only" selects 50% of the rows, if it is nicely packed that will probably be able to skip almost 50% of the pages and lead to efficient
But (especially when combining multiple filters), chances are the filter gets too fragmented, a 50% packed one is better than a 25% selective one that just does select 1 / skip 1 / select 1 ...
I think you could go even go further and check which combination of filtera are giving the best of both worlds (selectivity and fragmentation)
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "adriangbot (via GitHub)" <gi...@apache.org>.
adriangbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095989792
🤖 Arrow criterion benchmark running (GKE) | [trigger](https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095977120)
`Linux bench-c4095977120-475-64bv9 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux`
Comparing selectivity_threshold (5f2be6e96e561d86d4d496c2a6a953175a568901) to d53df60 (merge-base) [diff](https://github.com/apache/arrow-rs/compare/d53df605656d8012eca42e8ddffe165362a1a4cb..5f2be6e96e561d86d4d496c2a6a953175a568901)
BENCH_NAME=arrow_reader_clickbench
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench arrow_reader_clickbench
BENCH_FILTER=
Results will be posted here when complete
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-4095953829
I've some ideas to improve this further
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "alamb-ghbot (via GitHub)" <gi...@apache.org>.
alamb-ghbot commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3915641354
🤖 `./gh_compare_arrow.sh` [gh_compare_arrow.sh](https://github.com/alamb/datafusion-benchmarking/blob/main/scripts/gh_compare_arrow.sh) Running
Linux aal-dev 6.14.0-1018-gcp #19~24.04.1-Ubuntu SMP Wed Sep 24 23:23:09 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
Comparing selectivity_threshold (06954d51e887b073e37bdddfdb7a3ef83f9e7ca2) to df635903108418d95f7d0fc2101091684d8504fd [diff](https://github.com/apache/arrow-rs/compare/df635903108418d95f7d0fc2101091684d8504fd..06954d51e887b073e37bdddfdb7a3ef83f9e7ca2)
BENCH_NAME=arrow_reader_clickbench
BENCH_COMMAND=cargo bench --features=arrow,async,test_common,experimental,object_store --bench arrow_reader_clickbench
BENCH_FILTER=
BENCH_BRANCH_NAME=selectivity_threshold
Results will be posted here when complete
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Filter pushdown selectivity threshold [arrow-rs]
Posted by "Dandandan (via GitHub)" <gi...@apache.org>.
Dandandan commented on PR #9414:
URL: https://github.com/apache/arrow-rs/pull/9414#issuecomment-3907290665
@alamb can you get the runner "unstuck" again? :D
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org