feat: Use NestedLoopJoin instead of HashJoin/SortMergeJoin for small tables
#16450
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
github-actions
bot
added
core
|
I will try to run a benchmark on a table with smaller rows and return the result when finished. |
jonathanc-n
changed the title
NestedLoopJoinNestedLoopJoin
jonathanc-n
commented
Jun 18, 2025
| // If we can use nested loop join then we will combine the expressions in `join_on` | ||
| // and pass it into the join filter; create your join filters normally otherwise. | ||
| let join_filter: Option<JoinFilter> = if use_nested_loop_join_equijoin { | ||
| let combined_join_on_expression: Expr = filter |
There was a problem hiding this comment.
It seems that when we combine the JoinOn expressions here it will cause an error when both sides in the expression have the same unqualified name leading to duplicate unqualified fields. Is there a function that is able to qualify it with the schema, I can't seem to find one.
There was a problem hiding this comment.
To clarify, this is originally not a problem when the expression is in JoinOn as each unqualified column is referring to their own table before being combined into one expression.
|
The upside is that it performs well when both tables are extremely small < 50 rows 😆 |
|
Maybe it'll be clear to change the title as "Use NestedLoopJoin instead of HashJoin/SortMergeJoin for small tables", I'm confused when I saw the PR title first. |
Do we have some benchmark results? |
jonathanc-n
changed the title
NestedLoopJoinNestedLoopJoin instead of HashJoin/SortMergeJoin for small tables
|
@xudong963 These were tests run with one of the sides having 5 rows: Click to expandInteresting I noticed I wasn't using a filter for the The reason why |
|
I'll try to open a pull request later for creating a performance bench file for specifically benchmarking joins. |
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.
Which issue does this PR close?
Rationale for this change
We want to support equijoins in
NestedLoopJoinin the case where one of the tables in the join is very small.What changes are included in this PR?
I have added a
nested_loop_equijoin_thresholdto theOptimizerOptionswhich has a default value of 5 (same as DuckDB, here). This is the threshold for the number of rows that can be in either table so that the physical planner will choose aNestedLoopJoinExecoverSortMergeJoinandHashJoin.If either table has less than 5 rows then we will pass the
join_onexpressions to thejoin_filter.Are these changes tested?
By existing tests