[GLUTEN-1434] [VL] Fix shuffle read unalign buffer which origins from netty #2159
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
|
Run Gluten Clickhouse CI |
|
@jinchengchenghh can you please make a short notes in the PR description? -yuan |
jinchengchenghh
force-pushed
the
fixshuffle
branch
from
318a06f to
700df6b
Compare
|
Run Gluten Clickhouse CI |
zhztheplayer
approved these changes
Jul 3, 2023
zhztheplayer
reviewed
Jul 3, 2023
| values = convertToVeloxBuffer(arrowValueBuffer); | ||
| } else { | ||
| values = AlignedBuffer::allocate<char>(arrowValueBuffer->size(), pool); | ||
| memcpy(values->asMutable<char>(), arrowValueBuffer->data(), arrowValueBuffer->size()); |
There was a problem hiding this comment.
Is it possible to tweak on Java-side Netty library usages to make them conform to 16 byte alignment? So that we can avoid memcopy fundamentally
jinchengchenghh
force-pushed
the
fixshuffle
branch
from
700df6b to
32fc431
Compare
|
Run Gluten Clickhouse CI |
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Outdated suggestions cannot be applied.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.
For small size ColumnarBatch, this batch will not be compressed, the buffer which origins from netty is aligned, but the actual buffer used in RecordBatch is SliceBuffer(buffer, offset, size), this buffer cannot guarantee align. SIMD instruction
movdqarequired the address 16B aligned, so it will core dump at velox functioncopyValuesAndNullsand left potential coredump. This copy is expensive but essential.For most batch which is not tiny batch and compress codec use default lz4, shuffle read will decompress the buffer to an aligned address which meets SIMD instrictions requirement.
Relevant issue: facebookincubator/velox#2388