[
https://issues.apache.org/jira/browse/PIG-1443?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Daniel Dai updated PIG-1443:
----------------------------
Status: Resolved (was: Patch Available)
Hadoop Flags: [Reviewed]
Resolution: Fixed
Patch committed to both trunk and 0.7 branch.
> DefaultTuple underestimate the memory footprint for string
> ----------------------------------------------------------
>
> Key: PIG-1443
> URL: https://issues.apache.org/jira/browse/PIG-1443
> Project: Pig
> Issue Type: Bug
> Components: impl
> Affects Versions: 0.7.0
> Reporter: Daniel Dai
> Assignee: Daniel Dai
> Fix For: 0.7.0
>
> Attachments: PIG-1443-1.patch, PIG-1443-2.patch
>
>
> Currently, in DefaultTuple, we estimate the memory footprint for string as if
> it is char array. The formula we use is: length * 2 + 12. It turns out we
> underestimate the memory usage for string. Here is a list of real memory
> footprint for string we get from memory dump:
> | length of string | memory in bytes |
> | 7 | 56 |
> | 3 | 48 |
> | 1 | 40 |
> I did a search and find the following formula can accurately estimate the
> memory footprint for string:
> {code}
> 8 * (int) (((length * 2) + 45) / 8)
> {code}
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
