BryanCutler commented on issue #24734: [SPARK-27870][SQL][PySpark] Flush each batch for pandas UDF (for improving pandas UDFs pipeline) URL: https://github.com/apache/spark/pull/24734#issuecomment-497391057 What about sending large batches to Python, then taking smaller slices of each batch before sending for inference? Pyarrow supports zero-copy slices. In your use case you make 100 write calls to Python where 1 would be enough, and that adds some overhead.
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
