2010YOUY01 commented on code in PR #15700:
URL: https://github.com/apache/datafusion/pull/15700#discussion_r2191875936
##########
datafusion/physical-plan/src/aggregates/row_hash.rs:
##########
@@ -1067,14 +1074,13 @@ impl GroupedHashAggregateStream {
sort_batch(&batch, &expr, None)
})),
)));
- for spill in self.spill_state.spills.drain(..) {
- let stream =
self.spill_state.spill_manager.read_spill_as_stream(spill)?;
- streams.push(stream);
- }
+
self.spill_state.is_stream_merging = true;
self.input = StreamingMergeBuilder::new()
.with_streams(streams)
.with_schema(schema)
+ .with_spill_manager(self.spill_state.spill_manager.clone())
+ .with_sorted_spill_files(std::mem::take(&mut
self.spill_state.spills))
Review Comment:
Yes, but I think this is likely a small slowdown.
Comparing to 90% of the memory is taken by in-memory streams, and we have to
use the remaining 10% to re-spill lots of spills, this case can cause huge
slowdown.
--
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]
