arrow-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wes McKinney (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ARROW-110) [C++] Decide on optimal growth factor when appending to buffers/arrays
Date Sun, 08 Jan 2017 18:49:58 GMT

    [ https://issues.apache.org/jira/browse/ARROW-110?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15809840#comment-15809840
] 

Wes McKinney commented on ARROW-110:
------------------------------------

for arrays with 128K elements and less it uses 1.5; we could potentially implement a similar
strategy, once we write enough benchmarks to measure an impact on making lots of medium-ish
arrays https://github.com/facebook/folly/blob/master/folly/FBVector.h#L1141

> [C++] Decide on optimal growth factor when appending to buffers/arrays
> ----------------------------------------------------------------------
>
>                 Key: ARROW-110
>                 URL: https://issues.apache.org/jira/browse/ARROW-110
>             Project: Apache Arrow
>          Issue Type: Bug
>          Components: C++
>            Reporter: Micah Kornfield
>
> There is some evidence that powers of 2 might not be optimal (the facebook folly library
suggests this in there explanation of why they have there own vector type).  They use 1.5
(as do other implementations that don't use two).



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message