feat: stats pruning for LIKE filters #6049
Merged
Conversation
This file contains hidden or 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
Merging this PR will degrade performance by 40.14%
|
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | WallTime | u16_FoR[10M] |
6.5 µs | 10.8 µs | -40.14% |
Comparing aduffy/like-pushdown (e35f51d) with develop (a095ca7)
Footnotes
-
1290 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩
a10y
force-pushed
the
aduffy/like-pushdown
branch
2 times, most recently
from
8f4900b to
51f3fa7
Compare
github-actions
bot
removed
the
action/benchmark-sql
a10y
force-pushed
the
aduffy/like-pushdown
branch
from
51f3fa7 to
194c964
Compare
Codecov Report❌ Patch coverage is
☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
a10y
force-pushed
the
aduffy/like-pushdown
branch
from
e631ebd to
fbff6c0
Compare
Contributor
|
FWIW the stats truncation logic has the increment utf8 behaviour you want. It's implemented on a scalar https://github.com/vortex-data/vortex/blob/develop/vortex-scalar/src/binary.rs#L101 so it can mutate it in place. I wonder if they just have to stay separate but maybe upper bound being in place for binary scalar is just a cute optimistaion |
github-actions
bot
removed
the
action/benchmark-sql
Contributor
Benchmarks: TPC-H SF=1 on NVMESummary
Detailed Results Table
|
Contributor
Benchmarks: FineWeb NVMeSummary
Detailed Results Table
|
Contributor
Benchmarks: TPC-H SF=1 on S3Summary
Detailed Results Table
|
Contributor
Benchmarks: TPC-DS SF=1 on NVMESummary
Detailed Results Table
|
Contributor
Benchmarks: TPC-H SF=10 on NVMESummary
Detailed Results Table
|
Contributor
Benchmarks: FineWeb S3Summary
Detailed Results Table
|
Contributor
Benchmarks: Statistical and Population GeneticsSummary
Detailed Results Table
|
Contributor
Benchmarks: TPC-H SF=10 on S3Summary
Detailed Results Table
|
Contributor
Benchmarks: Clickbench on NVMESummary
Detailed Results Table
|
Contributor
Author
|
That function is missing some of the UTF-8 special character handling in the DF implementation, so now I'm worried that one of us is doing the wrong thing... |
Contributor
|
I gave you the binary version of this function, the utf8 version is here https://github.com/vortex-data/vortex/blob/develop/vortex-scalar/src/utf8.rs#L98 apologies |
Contributor
Author
|
I was looking at the utf8 version, I guess I don't know enough about utf8 to know if this covers the same edge cases that the datafusion implementation does |
Contributor
|
I derived it from the same code so we should make sure it’s the same so stats behave the same as regular filters |
a10y
force-pushed
the
aduffy/like-pushdown
branch
from
3927c66 to
c24c047
Compare
a10y
commented
Jan 23, 2026
| // Attempt to do min/max pruning for LIKE 'exact' or LIKE 'prefix%' | ||
|
|
||
| // Don't attempt to handle ilike or negated like | ||
| if like_opts.negated || like_opts.case_insensitive { |
Contributor
Author
There was a problem hiding this comment.
In theory we could handle negated like if we know that the block ONLY contains things which match the prefix, but we'd need to add a starts_with expression and I didn't want to do that here
Contributor
Author
There was a problem hiding this comment.
actually, we can do this without a starts_with, if we see that min >= prefix AND min < succ AND max >= prefix AND max < succ AND null_count = 0
Signed-off-by: Andrew Duffy <[email protected]>
Signed-off-by: Andrew Duffy <[email protected]>
Signed-off-by: Andrew Duffy <[email protected]>
a10y
force-pushed
the
aduffy/like-pushdown
branch
from
2e47faf to
e35f51d
Compare
robert3005
approved these changes
Jan 26, 2026
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 #6128
There are two types of
LIKEfilters that we can prune using our zone maps:col LIKE 'exact'col.min > 'exact' OR col.max < 'exact'col LIKE 'prefix%'col.min >= 'prefiy' OR col.max < 'prefix'I extracted out the existing logic from Utf8Scalar::upper_bound into a trait so we can make sure logic is consistent across stats gen/pruning