You are viewing a plain text version of this content. The canonical link for it is here.

Posted to jira@arrow.apache.org by "Yibo Cai (Jira)" <ji...@apache.org> on 2021/10/09 07:49:00 UTC

[jira] [Created] (ARROW-14269) [C++] Consolidate utf8 benchmark

Yibo Cai created ARROW-14269:
--------------------------------

             Summary: [C++] Consolidate utf8 benchmark
                 Key: ARROW-14269
                 URL: https://issues.apache.org/jira/browse/ARROW-14269
             Project: Apache Arrow
          Issue Type: Improvement
          Components: C++
            Reporter: Yibo Cai
            Assignee: Yibo Cai


I find some trivial (and obviously irrelevant) changes to UTF8 validation code may cause big variances to benchmark result.
UTF8 validation functions are inlined and called directly in benchmark. The compiler may try to optimize them together with the benchmark loop.
Un-inline the benchmark-ed functions makes the result predictable and explainable.



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