jimczi commented on PR #14076:
URL: https://github.com/apache/lucene/pull/14076#issuecomment-2563503703

   Thanks for looking @shatejas
   
   > The 
https://github.com/opensearch-project/k-NN/issues/2134#issuecomment-2420793541 
seemed to suggest it picked the last advice. I understand the setup wasn't 
constrained, would you be able to share the details of constrained setup where 
this caused unpredictable behavior. Just curious
   
   I may have overstated my point, I didn’t test it directly. My main concern 
is that the advice applies for the entire duration of the merge. Since the 
vector copy occurs at the start and represents only a small portion of the 
total merge time, I believe using prefetch instead of modifying the advice 
would be more suitable. Prefetch would allow us to fully control the readahead 
behavior. 
   
   > Are we sure that introducing prefetch during merge won't impact search?
   
   It would, especially when merging big segments, but only for the time of the 
copy, which should be relatively fast, rather than for the entire duration of 
the merge.


-- 
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: issues-unsubscr...@lucene.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org

Reply via email to