imay commented on issue #1236: Bad performance when query a table which has 
many columns(100+)
URL: 
https://github.com/apache/incubator-doris/issues/1236#issuecomment-498170972
 
 
   When I looked into code, I found the only difference between these two query 
is that in wide table, we will create a wide RowBlock for it, and these may 
lead to a wide memory range. And data in row block is arranged row by row, 
which means data in some column is not close. And this is bad for CPU cache.
   
   So I change the RowBlock's implementation in that RowBlock will only reserve 
memory for columns that will be used in query not in table schema. After this 
change, the `BlockConvertTime` reduce a lot. For the query for wide table 
(`table1`), the profile becomes like the following picture. 
   
   
![image](https://user-images.githubusercontent.com/1249159/58789121-f68c2a80-861f-11e9-9ccd-ce3e4bf51f11.png)
   
   So, I think the problem has been fixed. and I will push this patch later.

----------------------------------------------------------------
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]


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to