You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@arrow.apache.org by GitBox <gi...@apache.org> on 2022/10/14 04:19:31 UTC
[GitHub] [arrow-rs] tustvold opened a new issue, #2871: Deprecate Lexicographic Kernels
tustvold opened a new issue, #2871:
URL: https://github.com/apache/arrow-rs/issues/2871
**Is your feature request related to a problem or challenge? Please describe what you are trying to do.**
<!--
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
(This section helps Arrow developers understand the context and *why* for this feature, in addition to the *what*)
-->
`lexsort`, `lexsort_to_indices`, `lexicographical_partition_ranges`, etc... make use of `LexicographicalComparator` to compare rows. The branching and dynamic dispatch involved in this comparator is relatively expensive. Converting to the row format first, and comparing these rows has been found to offer significant performance advantages in similar applications - https://github.com/apache/arrow-datafusion/pull/3386.
**Describe the solution you'd like**
<!--
A clear and concise description of what you want to happen.
-->
We should provide examples, and potentially some utilities if necessary, to use the row format for these use-cases instead. We can then deprecate the lexicographic kernels and eventually remove them
**Describe alternatives you've considered**
<!--
A clear and concise description of any alternative solutions or features you've considered.
-->
We could leverage the row format within the existing kernels, however, this has a few drawbacks:
* The additional memory consumption is not necessarily obvious
* When chaining operators together, e.g. sort then partition, the row format conversion will be performed twice
* It isn't obvious how process data in batches
**Additional context**
<!--
Add any other context or screenshots about the feature request here.
-->
--
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.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-rs] comphead commented on issue #2871: Replace Lexicographic Kernels With Row Format
Posted by "comphead (via GitHub)" <gi...@apache.org>.
comphead commented on issue #2871:
URL: https://github.com/apache/arrow-rs/issues/2871#issuecomment-1645920427
@tustvold appreciate any details on how to use row format in sortExec
--
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
[GitHub] [arrow-rs] comphead commented on issue #2871: Replace Lexicographic Kernels With Row Format
Posted by "comphead (via GitHub)" <gi...@apache.org>.
comphead commented on issue #2871:
URL: https://github.com/apache/arrow-rs/issues/2871#issuecomment-1645957186
> > @tustvold appreciate any details on how to use row format in sortExec
>
> This was already implemented in [apache/arrow-datafusion#6163](https://github.com/apache/arrow-datafusion/pull/6163)
>
> I will file an upstream ticket for removing the last use of lexsort in DataFusion
Thanks please link the upstream ticket, I'd like to contribute on that one, as lexsort affects our use case.
--
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
[GitHub] [arrow-rs] tustvold commented on issue #2871: Replace Lexicographic Kernels With Row Format
Posted by "tustvold (via GitHub)" <gi...@apache.org>.
tustvold commented on issue #2871:
URL: https://github.com/apache/arrow-rs/issues/2871#issuecomment-1645928356
> @tustvold appreciate any details on how to use row format in sortExec
This was already implemented in https://github.com/apache/arrow-datafusion/pull/6163
--
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
[GitHub] [arrow-rs] Ted-Jiang commented on issue #2871: Replace Lexicographic Kernels With Row Format
Posted by GitBox <gi...@apache.org>.
Ted-Jiang commented on issue #2871:
URL: https://github.com/apache/arrow-rs/issues/2871#issuecomment-1290085228
I just wonder why not use row format in `sortExec`? 🤔 I am interesting about this! if there need any help, i would like to do😊
--
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
[GitHub] [arrow-rs] tustvold commented on issue #2871: Replace Lexicographic Kernels With Row Format
Posted by GitBox <gi...@apache.org>.
tustvold commented on issue #2871:
URL: https://github.com/apache/arrow-rs/issues/2871#issuecomment-1290106139
If you wanted to make a start migrating DataFusion's SortExec over to using the row format that would be amazing.
I'm currently working on some benchmarks for lexsort vs row format and I fully expect this to turn up some performance issues to fix.
If you are able to start migrating DF over in parallel, that would be awesome
--
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
[GitHub] [arrow-rs] tustvold commented on issue #2871: Replace Lexicographic Kernels With Row Format
Posted by "tustvold (via GitHub)" <gi...@apache.org>.
tustvold commented on issue #2871:
URL: https://github.com/apache/arrow-rs/issues/2871#issuecomment-1645959605
I have filed https://github.com/apache/arrow-datafusion/issues/7053 but I would strongly discourage you from picking it up, there is a recent I avoided doing it when I last played with SortExec, it is very fiddly to do well
--
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
[GitHub] [arrow-rs] tustvold closed issue #2871: Replace Lexicographic Kernels With Row Format
Posted by "tustvold (via GitHub)" <gi...@apache.org>.
tustvold closed issue #2871: Replace Lexicographic Kernels With Row Format
URL: https://github.com/apache/arrow-rs/issues/2871
--
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