You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Edward Capriolo (JIRA)" <ji...@apache.org> on 2013/10/22 03:05:41 UTC

[jira] [Commented] (HIVE-5602) Micro optimize select operator

    [ https://issues.apache.org/jira/browse/HIVE-5602?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13801368#comment-13801368 ] 

Edward Capriolo commented on HIVE-5602:
---------------------------------------

SELECT operator is doing try catch inside a for loop each column when it does not need to. Additionally we are making a function call each row to check conf.isSelectComputeNoStart()

I micro-benched before and after the change and showed a minimal bonus, please review.
{pre}
13/10/21 20:29:29 INFO exec.FilterOperator: 0 forwarding 1 rows
13/10/21 20:29:29 INFO exec.FilterOperator: 0 forwarding 10 rows
13/10/21 20:29:29 INFO exec.FilterOperator: 0 forwarding 100 rows
13/10/21 20:29:29 INFO exec.FilterOperator: 0 forwarding 1000 rows
13/10/21 20:29:29 INFO exec.FilterOperator: 0 forwarding 10000 rows
13/10/21 20:29:30 INFO exec.FilterOperator: 0 forwarding 100000 rows
13/10/21 20:29:31 INFO exec.FilterOperator: 0 forwarding 1000000 rows
13/10/21 20:29:33 INFO exec.FilterOperator: 0 forwarding 2000000 rows
13/10/21 20:29:34 INFO exec.FilterOperator: 0 forwarding 3000000 rows
13/10/21 20:29:36 INFO exec.FilterOperator: 0 forwarding 4000000 rows
13/10/21 20:29:38 INFO exec.FilterOperator: 0 forwarding 5000000 rows
13/10/21 20:29:40 INFO exec.FilterOperator: 0 forwarding 6000000 rows
13/10/21 20:29:41 INFO exec.FilterOperator: 0 forwarding 7000000 rows
13/10/21 20:29:43 INFO exec.FilterOperator: 0 forwarding 8000000 rows
13/10/21 20:29:45 INFO exec.FilterOperator: 0 forwarding 9000000 rows
13/10/21 20:29:46 INFO exec.FilterOperator: 0 forwarding 10000000 rows

13/10/21 20:31:36 INFO exec.FilterOperator: Initialization Done 0 FIL
13/10/21 20:31:36 INFO exec.FilterOperator: 0 forwarding 1 rows
13/10/21 20:31:36 INFO exec.FilterOperator: 0 forwarding 10 rows
13/10/21 20:31:36 INFO exec.FilterOperator: 0 forwarding 100 rows
13/10/21 20:31:36 INFO exec.FilterOperator: 0 forwarding 1000 rows
13/10/21 20:31:37 INFO exec.FilterOperator: 0 forwarding 10000 rows
13/10/21 20:31:37 INFO exec.FilterOperator: 0 forwarding 100000 rows
13/10/21 20:31:38 INFO exec.FilterOperator: 0 forwarding 1000000 rows
13/10/21 20:31:40 INFO exec.FilterOperator: 0 forwarding 2000000 rows
13/10/21 20:31:41 INFO exec.FilterOperator: 0 forwarding 3000000 rows
13/10/21 20:31:43 INFO exec.FilterOperator: 0 forwarding 4000000 rows
13/10/21 20:31:45 INFO exec.FilterOperator: 0 forwarding 5000000 rows
13/10/21 20:31:46 INFO exec.FilterOperator: 0 forwarding 6000000 rows
13/10/21 20:31:48 INFO exec.FilterOperator: 0 forwarding 7000000 rows
13/10/21 20:31:49 INFO exec.FilterOperator: 0 forwarding 8000000 rows
13/10/21 20:31:51 INFO exec.FilterOperator: 0 forwarding 9000000 rows
13/10/21 20:31:53 INFO exec.FilterOperator: 0 forwarding 10000000 rows
{pre}

> Micro optimize select operator
> ------------------------------
>
>                 Key: HIVE-5602
>                 URL: https://issues.apache.org/jira/browse/HIVE-5602
>             Project: Hive
>          Issue Type: Improvement
>            Reporter: Edward Capriolo
>            Assignee: Edward Capriolo
>            Priority: Minor
>         Attachments: HIVE-5602.patch.1.txt
>
>




--
This message was sent by Atlassian JIRA
(v6.1#6144)