Fix reserialization of broadcasted tables #3504
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
Signed-off-by: Jason Lowe <jlowe@nvidia.com>
|
build |
|
build |
1 similar comment
|
build |
revans2
previously approved these changes
Sep 16, 2021
|
build |
|
looks like a lot of test failures and might be related to changes |
|
Test failures were related to row-count-only tables. I updated to match the old behavior where we don't try to serialize or deserialize the types if there's only rows. I could simplify this code by serializing the datatypes in all cases to avoid the special-case scenarios at the cost of a bit bigger serialization data for the row-count-only cases. Let me know if that's desirable. |
|
build |
revans2
approved these changes
Sep 20, 2021
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Outdated suggestions cannot be applied.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.
Fixes #3266.
Re-serializing a broadcast caused issues because it purposely deserializes into a different state on the executor than it does on the driver, so serializing on the executor due to memory pressure after being deserialized could cause an NPE.
This updates the broadcast to always deserialize only to memory but then throw away the host buffers once the GPU batch is requested at which point it creates it.