[GitHub] [druid] jasonk000 commented on pull request #12302: perf: core TextReader use a smaller default buffer

Mon, 07 Nov 2022 17:05:31 -0800 


jasonk000 commented on PR #12302:
URL: https://github.com/apache/druid/pull/12302#issuecomment-1306458118

   @abhishekagarwal87 -- as requested, some flamegraphs here that give a bit 
more detail. I've spun up a half-cluster to try to get a clean a/b overview 
between the nodes but that's a little more tricky.
   
   BEFORE
   15.6% of Java CPU is in `TextReader`
   
![image](https://user-images.githubusercontent.com/3196528/200445748-9fc70f79-474b-4a66-8f58-86ee2ab03942.png)
   
   AFTER
   10.7% of Java CPU is in `TextReader`
   
![image](https://user-images.githubusercontent.com/3196528/200445692-cf87ab6b-ae3e-44a8-9ad4-59c63ac6a811.png)
   
   ZOOMING IN
   - zooming in to one of the `StreamChunkParser` paths
   - highlighting `BufferedReader.*ini`
   - note that the reader init code disappears completely
   
   BEFORE
   
   
![image](https://user-images.githubusercontent.com/3196528/200446915-991f67a8-d0cb-4b65-84e0-514ab18d6243.png)
   
   AFTER
   
   
![image](https://user-images.githubusercontent.com/3196528/200446951-2f5d5ef4-4442-423b-91af-b150e8854d6d.png)
   
   


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


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

Reply via email to