[
https://issues.apache.org/jira/browse/CALCITE-3785?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17035227#comment-17035227
]
Stamatis Zampetakis commented on CALCITE-3785:
----------------------------------------------
[~hyuan] I took the liberty to include your findings and profiling screenshots
in this case.
> Performance regression in HepPlanner#belongsToDag() due to expensive hashCode
> computation
> -----------------------------------------------------------------------------------------
>
> Key: CALCITE-3785
> URL: https://issues.apache.org/jira/browse/CALCITE-3785
> Project: Calcite
> Issue Type: Bug
> Reporter: Xiening Dai
> Priority: Major
> Labels: pull-request-available
> Attachments: after.png, before.png
>
> Time Spent: 40m
> Remaining Estimate: 0h
>
> It turns out that the change in CALCITE-3713 added an extra overhead of
> 20~30% HepPlanner. The hash map lookup in belongsToDag becomes much slower
> due to expensive computation of the hashcode and equals for the list.
> A simple profiling of the planner before and after CALCITE-3713 is shown
> below.
> !before.png!
> !after.png!
> To test whether or not a vertex belongs to DAG, we can simply do
> graph.vertexSet().contains(vertex); There's no need to look up in
> mapDigestToVertex map, which incurs overhead of creating the map key. This
> problem was amplified by CALCITE-3713.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
