rluvaton commented on code in PR #19494:
URL: https://github.com/apache/datafusion/pull/19494#discussion_r2651137705
##########
datafusion/physical-plan/src/sorts/sort.rs:
##########
@@ -812,23 +827,24 @@ impl ExternalSorter {
}
}
-/// Estimate how much memory is needed to sort a `RecordBatch`.
-///
-/// This is used to pre-reserve memory for the sort/merge. The sort/merge
process involves
-/// creating sorted copies of sorted columns in record batches for speeding up
comparison
-/// in sorting and merging. The sorted copies are in either row format or
array format.
-/// Please refer to cursor.rs and stream.rs for more details. No matter what
format the
-/// sorted copies are, they will use more memory than the original record
batch.
-pub(crate) fn get_reserved_byte_for_record_batch_size(record_batch_size:
usize) -> usize {
- // 2x may not be enough for some cases, but it's a good start.
- // If 2x is not enough, user can set a larger value for
`sort_spill_reservation_bytes`
- // to compensate for the extra memory needed.
- record_batch_size * 2
+/// Calculate how much memory to reserve for sorting a `RecordBatch` from its
size,
+/// this can be calculated as the sum of the actual space it takes in
memory(which would be larger for a sliced batch),
+/// and the size of the actual data.
+pub(crate) fn get_reserved_byte_for_record_batch_size(
+ record_batch_size: usize,
+ sliced_size: usize,
+) -> usize {
+ record_batch_size + sliced_size
}
Review Comment:
The comments still are still valid, please keep them
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
