[Data][3/N] Enable optimizer: fix stats and RandomizeBlocks #35952
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
approved these changes
May 31, 2023
scottjlee
approved these changes
Jun 1, 2023
| # callable object. | ||
| return fn.__class__.__name__ | ||
| except AttributeError as e: | ||
| logging.error("Failed to get name of UDF %s: %s", fn, e) |
There was a problem hiding this comment.
should we use DatasetLogger here for consistency? https://github.com/ray-project/ray/blob/master/python/ray/data/_internal/dataset_logger.py
| @@ -374,7 +374,7 @@ def test_map_batches_basic(ray_start_regular_shared, tmp_path, restore_data_cont | |||
|
|
|||
| def test_map_batches_extra_args(shutdown_only, tmp_path): | |||
| ray.shutdown() | |||
| ray.init(num_cpus=2) | |||
| ray.init(num_cpus=3) | |||
There was a problem hiding this comment.
is this needed as a result of code changes in this PR?
There was a problem hiding this comment.
This is because we are no longer fusing actors.
scv119
pushed a commit
to scv119/ray
that referenced
this pull request
Jun 16, 2023
…ect#35952) ## Why are these changes needed? This is the last PR for enabling optimizer by default. It contains the following changes: - Fix DatasetStats related issues, including: - Map op names not including function names. - Read op names not including data source names. - RandomizeBlocks's name. - `generate_randomize_blocks_fn` returning empty stats, making it being skipped in summary. - Dropping support for fusing 2 actor-based map ops. - In the old backend, we fuse 2 actors only if they are the same class and have the same constructor args. This is not useful in practice. Test changes about `num_cpus` are related to this. - Fix the issue that `ReorderRandomizeBlocksRule` will modify the operator's input_dependencies in place. This is a bug because the operator instance might be shared by multiple Datasets. ## Related issue number Closes ray-project#32596
arvind-chandra
pushed a commit
to lmco/ray
that referenced
this pull request
Aug 31, 2023
…ect#35952) ## Why are these changes needed? This is the last PR for enabling optimizer by default. It contains the following changes: - Fix DatasetStats related issues, including: - Map op names not including function names. - Read op names not including data source names. - RandomizeBlocks's name. - `generate_randomize_blocks_fn` returning empty stats, making it being skipped in summary. - Dropping support for fusing 2 actor-based map ops. - In the old backend, we fuse 2 actors only if they are the same class and have the same constructor args. This is not useful in practice. Test changes about `num_cpus` are related to this. - Fix the issue that `ReorderRandomizeBlocksRule` will modify the operator's input_dependencies in place. This is a bug because the operator instance might be shared by multiple Datasets. ## Related issue number Closes ray-project#32596 Signed-off-by: e428265 <arvind.chandramouli@lmco.com>
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 is the last PR for enabling optimizer by default. It contains the following changes:
generate_randomize_blocks_fnreturning empty stats, making it being skipped in summary.num_cpusare related to this.ReorderRandomizeBlocksRulewill modify the operator's input_dependencies in place. This is a bug because the operator instance might be shared by multiple Datasets.Related issue number
Closes #32596
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.