steFaiz opened a new pull request, #8064:
URL: https://github.com/apache/paimon/pull/8064

   ### Purpose
   This PR will close: https://github.com/apache/paimon/pull/8010
   I tested a data-evolution table of 100,000,000 records, several structured 
columns and a blob column.
   The result is as below:
   
   | Metrics | Value |
   | --- | --- |
   | Plan | 49.78s |
   | **AVG Per chunk read** | **1.199s** |
   | chunk size | 100 |
   | AVG chunk Arrow size | 41.14 MiB |
   | AVG chunk file num | 81 |
   | columns | length, image_name, conversations, width, height, image_count, 
dataset, image_bytes |
   
   I directly test reading tables on dfs, it costs a lot to plan i.e. generate 
1 million DataSplits and shuffle them. This is because generating 1 million 
objects in Python is heavy. This will be completed within several hundred of 
millisecond is Java.
   
   Next step I'll try to add shuffle and buffered shuffle for Pytorch Paimon 
Dataset.
   
   ### Tests
   See Unit Tests


-- 
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