milenkovicm commented on code in PR #8004:
URL: https://github.com/apache/arrow-datafusion/pull/8004#discussion_r1377737703
##########
datafusion/physical-plan/src/aggregates/row_hash.rs:
##########
@@ -673,7 +673,18 @@ impl GroupedHashAggregateStream {
let spillfile =
self.runtime.disk_manager.create_tmp_file("HashAggSpill")?;
let mut writer = IPCWriter::new(spillfile.path(), &emit.schema())?;
// TODO: slice large `sorted` and write to multiple files in parallel
- writer.write(&sorted)?;
+ let mut offset = 0;
+ let total_rows = sorted.num_rows();
+
+ while offset < total_rows {
+ // TODO: we could consider smaller batch size as there may be
hundreds of batches
Review Comment:
There is a typo "...batch size as there may be hundreds of batches" ->
"batch size as there may be hundreds of FILES"
What I meant if we have more files spilled, it makes sense, in my head at
least, to have smaller batch size as would decrease memory pressure.
--
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]
