Re: [PR] [CELEBORN-1931] use gather API for local flusher to optimize write io pattern [celeborn]

Tue, 08 Apr 2025 01:39:57 -0700 


FMX commented on PR #3178:
URL: https://github.com/apache/celeborn/pull/3178#issuecomment-2785676504

   I saw some performance improvements for this PR. 
   
   For a test job with 300GB shuffle size, the writer time reduced from 10 min 
to 9.7 min.
   
   Enable Gather API 
   
![lQLPJxl8kDYuZwdmzQlKsHCuYYMCT188B9bA0PE-EgA_2378_102](https://github.com/user-attachments/assets/7999a334-2dd2-4d48-bc74-3642b1a7e850)
   
![lQLPKH431fi5FwfNAnTNBuSww5YEPrszAEcH1sDQ8X-gAQ_1764_628](https://github.com/user-attachments/assets/b702115a-039c-41c6-900b-5cc690ce0f60)
   
   
   Disable Gather API
   
![lQLPJwkbyvHERwdQzQkosIyk8KdqmezQB9bA0PE-EgE_2344_80](https://github.com/user-attachments/assets/8c327ff4-8bee-49a0-8bd5-81fa4e300e28)
   
![lQLPJwWOTEz9xwfNArjNBuqwd9rVxayjEicH1sDQ8X-gAA_1770_696](https://github.com/user-attachments/assets/8bc4a476-2793-44b7-8479-be777995fac8)
   
   


-- 
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