[
https://issues.apache.org/jira/browse/ARROW-13923?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17448938#comment-17448938
]
Yibo Cai commented on ARROW-13923:
----------------------------------
Not satisfied with my current simd implementation.
It's too complex. It can only handle FindLast, not FindNth. And it performs bad
if there are continuous escapes (\\\\....).
https://github.com/apache/arrow/pull/11101 looks a better approach.
> [C++] Improve CSV chunker with SIMD
> -----------------------------------
>
> Key: ARROW-13923
> URL: https://issues.apache.org/jira/browse/ARROW-13923
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
> Reporter: Yibo Cai
> Assignee: Yibo Cai
> Priority: Major
> Attachments: 1.diff
>
>
> POC test shows about 5x performance improvement when leveraging simd to
> optimize csv boundary finder.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)