[jira] [Commented] (ARROW-11030) [Rust] [DataFusion] MutableArrayData slow with many batches

["Daniël Heres (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ARROW-11030?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17256109#comment-17256109 ] 

Daniël Heres commented on ARROW-11030:
--------------------------------------

One comment I put in a PR, which is I think part of the problem:

I think part of a further speed up could be moving the building of the left / build-side Vec<&ArrayData> arrays so that it is only created once instead of for each right batch in build_batch_from_indices. Currently when making the batch size smaller, the build-side Vec is built more times, but also contains more (smaller) batches itself, which could explain (part of the) big / exponential slowdown on smaller batches.

> [Rust] [DataFusion] MutableArrayData slow with many batches
> -----------------------------------------------------------
>
>                 Key: ARROW-11030
>                 URL: https://issues.apache.org/jira/browse/ARROW-11030
>             Project: Apache Arrow
>          Issue Type: Bug
>          Components: Rust - DataFusion
>            Reporter: Andy Grove
>            Assignee: Andy Grove
>            Priority: Major
>             Fix For: 3.0.0
>
>
> Performance of joins slows down dramatically with smaller batches.
> The issue is related to slow performance of MutableDataArray::new() when passed a high number of batches. This happens when passing in all of the batches from the build side of the join and this happens once per build-side join key for each probe-side batch.
> It seems to get exponentially slower as the number of arrays increases even though the number of rows is the same.
> I modified hash_join.rs to have this debug code:
> {code:java}
> let start = Instant::now();
> let row_count: usize = arrays.iter().map(|arr| arr.len()).sum();
> let num_arrays = arrays.len();
> let mut mutable = MutableArrayData::new(arrays, true, capacity);
> if num_arrays > 0 {
>     debug!("MutableArrayData::new() with {} arrays containing {} rows took {} ms", num_arrays, row_count, start.elapsed().as_millis());
> } {code}
> Batch size 131072:
> {code:java}
> MutableArrayData::new() with 4584 arrays containing 3115341 rows took 1 ms
> MutableArrayData::new() with 4584 arrays containing 3115341 rows took 1 ms
> MutableArrayData::new() with 4584 arrays containing 3115341 rows took 1 ms {code}
> Batch size 16384:
> {code:java}
> MutableArrayData::new() with 36624 arrays containing 3115341 rows took 19 ms
> MutableArrayData::new() with 36624 arrays containing 3115341 rows took 16 ms
> MutableArrayData::new() with 36624 arrays containing 3115341 rows took 17 ms {code}
> Batch size 4096:
> {code:java}
> MutableArrayData::new() with 146496 arrays containing 3115341 rows took 88 ms
> MutableArrayData::new() with 146496 arrays containing 3115341 rows took 89 ms
> MutableArrayData::new() with 146496 arrays containing 3115341 rows took 88 ms {code}
>  
>  
>  
>  
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)