Refactor batch coalesce to be based solely on batch data size #1133
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
Signed-off-by: Jason Lowe <[email protected]>
|
build |
revans2
previously approved these changes
Nov 17, 2020
sql-plugin/src/main/scala/com/nvidia/spark/rapids/GpuCoalesceBatches.scala
Show resolved
Hide resolved
Signed-off-by: Jason Lowe <[email protected]>
|
build |
|
I checked with @nvdbaranec on cudf's concatenate and apparently only one of the two possible code paths is properly checking for a row count overflow. I'm marking this as a draft PR until a fix for that goes in, since this is removing the explicit string limit check in the plugin and relies on cudf::concatenate to catch this. |
|
This is blocked pending rapidsai/cudf#6809 |
|
The cudf dependency has been merged to cudf 0.17, so this is finally ready to go. |
|
build |
abellina
approved these changes
Dec 2, 2020
nartal1
pushed a commit
to nartal1/spark-rapids
that referenced
this pull request
Jun 9, 2021
…#1133) * Refactor batch coalesce to be based solely on batch data size Signed-off-by: Jason Lowe <[email protected]> * Add TargetSize limit check and comments Signed-off-by: Jason Lowe <[email protected]>
nartal1
pushed a commit
to nartal1/spark-rapids
that referenced
this pull request
Jun 9, 2021
…#1133) * Refactor batch coalesce to be based solely on batch data size Signed-off-by: Jason Lowe <[email protected]> * Add TargetSize limit check and comments Signed-off-by: Jason Lowe <[email protected]>
tgravescs
pushed a commit
to tgravescs/spark-rapids
that referenced
this pull request
Nov 30, 2023
…IDIA#1133) Signed-off-by: spark-rapids automation <[email protected]>
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.
#1116 placed a 2GB limit on the GPU batch target size which will prevent us from trying to coalesce batches that can exceed the 2GB element limit for any single column. This PR refactors batch coalesce so it is based solely on data size rather than needing to also track string sizes. This significantly simplifies the coalesce logic and also removes the need for the plugin to traverse the column hierarchy to calculate individual column sizes when the batch is single-buffer based (e.g.: compressed batch or
contiguous_splitbatch).Once cudf's
ColumnVector#getDeviceMemorySizeis updated to account for nested columns then even normal GPU batches only will need the plugin to traverse the top-level columns and not need to check for string/list types.Note that the original batch coalesce code checked for batches containing
RapidsHostColumnVector, but those column types should never be seen by a batch coalesce. These types of batches only exist specifically between a partition and the shuffle, and there should never be a coalesce in-between. This also removes the implicit size fallback if the column type is ignored.