alamb opened a new issue, #7023:
URL: https://github.com/apache/arrow-datafusion/issues/7023

   ### Is your feature request related to a problem or challenge?
   
   During review of https://github.com/apache/arrow-datafusion/pull/6932 there 
were several suggstions made about how we could improve the performance of 
streaming / bounded grouping (when the data is sorted by some/all of the group 
keys)
   
   This ticket tracks those improvements
   
   
   ### Describe the solution you'd like
   
   
   # Suggestion 1
   @tustvold  suggests on 
https://github.com/apache/arrow-datafusion/pull/6932#r1267305736
   
   > It occurs to me that a potentially faster way to detect the group 
boundaries would be to use the inequality kernel offset by one w.r.t to each 
other, ORing the results together, and then iterating over the set bits.
   > 
   > There would be some subtlety to handle nulls correctly, but it would 
likely be significantly faster and would not require converting to the row 
format
   >
   > We could likely do something similar for window functions if we aren't 
already
   
   Note that IOx does something similar to compute partition boundaries here: 
https://github.com/influxdata/influxdb_iox/blob/d3b3805e5fa87214ea934427df3967a57a14669c/iox_query/src/provider/deduplicate/algo.rs#L101-L207
   
   # Suggestion 2
   From https://github.com/apache/arrow-datafusion/pull/6932#r1267344573
   
   > In the case of GroupOrderingFull it seems unnecessary to be computing 
hashes at all, we can just group based on whenever the sort key changes?
   
   This could likely go quite a bit faster as hashing is a significant part of 
the time
   
   ### Describe alternatives you've considered
   
   _No response_
   
   ### Additional context
   
   _No response_


-- 
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]

Reply via email to