[data] Stage fusion optimizations, off by default #22373
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ericl
changed the title
clarkzinzow
requested changes
Feb 16, 2022
clarkzinzow
approved these changes
Feb 17, 2022
python/ray/data/tests/test_stats.py
Outdated
| @@ -167,6 +167,7 @@ def test_dataset_pipeline_stats_basic(ray_start_regular_shared): | |||
| for batch in pipe.iter_batches(): | |||
| pass | |||
| stats = canonicalize(pipe.stats()) | |||
| print(stats) | |||
There was a problem hiding this comment.
Suggested change
| print(stats) |
simonsays1980
pushed a commit
to simonsays1980/ray
that referenced
this pull request
Feb 27, 2022
This PR adds the following stage fusion optimizations (off by default). In a later PR, I plan to enable this by default for DatasetPipelines.
- Stage fusion: Whether to fuse compatible OneToOne stages.
- Read stage fusion: Whether to fuse read stages into downstream OneToOne stages. This is accomplished by rewriting the read stage (LazyBlockList) into a transformation over a collection of read tasks (BlockList -> MapBatches(do_read)).
- Shuffle stage fusion: Whether to fuse compatible OneToOne stages into shuffle stages that support specifying a map-side block UDF.
Stages are considered compatible if their compute strategy is the same ("tasks" vs "actors"), and they have the same Ray remote args. Currently, the PR is ignoring the remote args of read tasks, but this will be fixed as a followup (I didn't want to change the read tasks default here).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why are these changes needed?
This PR adds the following stage fusion optimizations (off by default). In a later PR, I plan to enable this by default for DatasetPipelines.
Stages are considered compatible if their compute strategy is the same ("tasks" vs "actors"), and they have the same Ray remote args. Currently, the PR is ignoring the remote args of read tasks, but this will be fixed as a followup (I didn't want to change the read tasks default here).
I've experimented with this locally and the memory reduction is ~50%, mostly from clearing input blocks. In the distributed setting I'd expect more savings especially under memory pressure, since fusion suppresses excess spilling / transfer of blocks.
Related issue number
Towards #18791