[
https://issues.apache.org/jira/browse/SPARK-13131?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15131353#comment-15131353
]
Sean Owen commented on SPARK-13131:
-----------------------------------
Isn't best time on fact the best estimator of what this benchmark seeks to
measure - how much pure computation X takes? Run time is true constant time
taken to compute X plus noise from other things executing concurrently. The
'error' can never be negative. Least run time is then the one with least error.
Mean isn't that helpful in comparison. Stability is a secondary issue
> Use best time and average time in micro benchmark
> --------------------------------------------------
>
> Key: SPARK-13131
> URL: https://issues.apache.org/jira/browse/SPARK-13131
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Reporter: Davies Liu
> Assignee: Davies Liu
>
> Best time should be more stable than average time in benchmark, together with
> average time, they could show more information.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
