[
https://issues.apache.org/jira/browse/HIVE-24510?focusedWorklogId=532560&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-532560
]
ASF GitHub Bot logged work on HIVE-24510:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 07/Jan/21 15:44
Start Date: 07/Jan/21 15:44
Worklog Time Spent: 10m
Work Description: abstractdog edited a comment on pull request #1824:
URL: https://github.com/apache/hive/pull/1824#issuecomment-756195387
ran through the patches, I'll comment on particular code sections, but first
a question in general: is there a specific reason for doing this method
transformation?
```
compute_bit_vector(x, 'hll') -> compute_bit_vector_hll(x)
```
while I was reading the qout changes I realized that this rewrite makes this
patch so huge, and I'm not sure if it's really worth, as I cannot see the added
value, and I personally don't have any problem with the algorithm as a
parameter, also what's the default behavior? this would make sense to me:
```
compute_bit_vector(x) -- could be translated to --> compute_bit_vector(x,
'hll')
```
without added value, e.g. the backport of this patch will be painful if we
keep it this way, and the patch is about vectorization coverage, and has
nothing to do with compiler basically, what do you think?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 532560)
Time Spent: 1h 10m (was: 1h)
> Vectorize compute_bit_vector
> ----------------------------
>
> Key: HIVE-24510
> URL: https://issues.apache.org/jira/browse/HIVE-24510
> Project: Hive
> Issue Type: Improvement
> Reporter: Mustafa İman
> Assignee: Mustafa İman
> Priority: Major
> Labels: pull-request-available
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> After https://issues.apache.org/jira/browse/HIVE-23530 , almost all compute
> stats functions are vectorizable. Only function that is not vectorizable is
> "compute_bit_vector" for ndv statistics computation. This causes "create
> table as select" and "insert overwrite select" queries to run in
> non-vectorized mode.
> Even a very naive implementation of vectorized compute_bit_vector gives about
> 50% performance improvement on simple "insert overwrite select" queries. That
> is because entire mapper or reducer can run in vectorized mode.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
