[opt](nereids) support partitionTopn for multi window exprs#38393
Merged
englefly merged 6 commits intoapache:masterfrom Aug 6, 2024
Merged
[opt](nereids) support partitionTopn for multi window exprs#38393englefly merged 6 commits intoapache:masterfrom
englefly merged 6 commits intoapache:masterfrom
Conversation
|
Thank you for your contribution to Apache Doris. Since 2024-03-18, the Document has been moved to doris-website. |
Contributor
Author
|
run buildall |
1 similar comment
Contributor
Author
|
run buildall |
xzj7019
force-pushed
the
support_multi_partitionTopn
branch
from
7e65c5a to
29f6a0b
Compare
Contributor
Author
|
run buildall |
xzj7019
force-pushed
the
support_multi_partitionTopn
branch
from
29f6a0b to
04ff675
Compare
Contributor
Author
|
run buildall |
TPC-H: Total hot run time: 39375 ms |
TPC-DS: Total hot run time: 172696 ms |
ClickBench: Total hot run time: 30.74 s |
Contributor
Author
|
run buildall |
TPC-H: Total hot run time: 39737 ms |
TPC-DS: Total hot run time: 172790 ms |
ClickBench: Total hot run time: 31.38 s |
Contributor
Author
|
run buildall |
TPC-H: Total hot run time: 39385 ms |
TPC-DS: Total hot run time: 173456 ms |
ClickBench: Total hot run time: 30.41 s |
englefly
reviewed
Jul 29, 2024
Contributor
|
please add a sql example in "Proposed changes" section |
Contributor
Author
Done |
zhangstar333
previously approved these changes
Jul 31, 2024
github-actions
bot
added
the
approved
Contributor
|
PR approved by at least one committer and no changes requested. |
Contributor
|
PR approved by anyone and no changes requested. |
morrySnow
reviewed
Aug 2, 2024
| * pushPartitionLimitThroughWindow is used to push the partitionLimit through the window | ||
| * and generate the partitionTopN. If the window can not meet the requirement, | ||
| * it will return null. So when we use this function, we need check the null in the outside. | ||
| * check and get valid window function and partition limit value |
Contributor
There was a problem hiding this comment.
add comment to explain the input parameter and return value
| long chosenRowNumberPartitionLimit = Long.MAX_VALUE; | ||
| boolean hasRowNumber = false; | ||
| for (NamedExpression windowExpr : windowExpressions) { | ||
| WindowExpression windowFunc = (WindowExpression) windowExpr.child(0); |
| Set<Expression> conjuncts = filter.getConjuncts(); | ||
| Set<Expression> relatedConjuncts = extractRelatedConjuncts(conjuncts, windowExpr.getExprId()); | ||
| for (Expression conjunct : relatedConjuncts) { | ||
| Preconditions.checkArgument(conjunct instanceof BinaryOperator); |
Contributor
There was a problem hiding this comment.
add error msg? log warning log and return null instead throw exception?
Contributor
Author
There was a problem hiding this comment.
remove the pre-condition checking since the former logic has been done.
Comment on lines
+251
to
+252
| Preconditions.checkArgument(leftChild instanceof SlotReference | ||
| && rightChild instanceof IntegerLikeLiteral); |
Contributor
There was a problem hiding this comment.
add error msg? log warning log and return null instead throw exception?
Contributor
Author
There was a problem hiding this comment.
remove the pre-condition checking since the former logic has been done.
xzj7019
force-pushed
the
support_multi_partitionTopn
branch
from
d053abc to
95c8ad6
Compare
Contributor
Author
|
run buildall |
TPC-H: Total hot run time: 41944 ms |
TPC-DS: Total hot run time: 168646 ms |
ClickBench: Total hot run time: 30.61 s |
englefly
approved these changes
Aug 6, 2024
github-actions
bot
added
the
approved
Contributor
|
PR approved by at least one committer and no changes requested. |
zhangstar333
approved these changes
Aug 6, 2024
dataroaring
pushed a commit
that referenced
this pull request
Aug 11, 2024
Support partitionTopn for multi window exprs. If row_number exists, choose row_number with the minimal limit value; if not, choose others with the minimal limit value. Example: ``` mysql> explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, rank() over(partition by c1 order by c3) as rk from push_down_multi_predicate_through_window_t) t where rn <= 1 and rk <= 1; +----------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +----------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rk <= 1) and (rn <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[LOCAL_SORT] | | ----------PhysicalDistribute[DistributionSpecHash] | | ------------PhysicalWindow | | --------------PhysicalQuickSort[LOCAL_SORT] | | ----------------PhysicalDistribute[DistributionSpecHash] | | ------------------PhysicalPartitionTopN | | --------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +----------------------------------------------------------------------------------+ ``` --------- Co-authored-by: zhongjian.xzj <zhongjian.xzj@zhongjianxzjdeMacBook-Pro.local>
morrySnow
pushed a commit
that referenced
this pull request
Aug 13, 2024
intro by #38393 Fix the cases whose window function both contains row_number and other types but only the other types contains pushing down filter.
wyxxxcat
pushed a commit
to wyxxxcat/doris
that referenced
this pull request
Aug 14, 2024
…e#39233) intro by apache#38393 Fix the cases whose window function both contains row_number and other types but only the other types contains pushing down filter.
dataroaring
pushed a commit
that referenced
this pull request
Aug 16, 2024
Support partitionTopn for multi window exprs. If row_number exists, choose row_number with the minimal limit value; if not, choose others with the minimal limit value. Example: ``` mysql> explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, rank() over(partition by c1 order by c3) as rk from push_down_multi_predicate_through_window_t) t where rn <= 1 and rk <= 1; +----------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +----------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rk <= 1) and (rn <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[LOCAL_SORT] | | ----------PhysicalDistribute[DistributionSpecHash] | | ------------PhysicalWindow | | --------------PhysicalQuickSort[LOCAL_SORT] | | ----------------PhysicalDistribute[DistributionSpecHash] | | ------------------PhysicalPartitionTopN | | --------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +----------------------------------------------------------------------------------+ ``` --------- Co-authored-by: zhongjian.xzj <zhongjian.xzj@zhongjianxzjdeMacBook-Pro.local>
dataroaring
pushed a commit
that referenced
this pull request
Aug 17, 2024
intro by #38393 Fix the cases whose window function both contains row_number and other types but only the other types contains pushing down filter.
yiguolei
pushed a commit
that referenced
this pull request
Aug 22, 2024
## Proposed changes pick from #38393 Co-authored-by: xiongzhongjian <xiongzhongjian@selectdb.com>
Closed
16 tasks
morrySnow
pushed a commit
that referenced
this pull request
Nov 27, 2024
…forbidden type (#44617) Related PR: #38393 Problem Summary: In the previous pr which supporting multi win expr ptopN pushdown, it handled partial forbidden type unexpectly and will lead some case to push down the pTopN wrongly. plan before fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +------------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +------------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalPartitionTopN | | ----------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +------------------------------------------------------------------------------------+ plan after fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +----------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +----------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +----------------------------------------------------------------------------------+
github-actions bot
pushed a commit
that referenced
this pull request
Nov 27, 2024
…forbidden type (#44617) Related PR: #38393 Problem Summary: In the previous pr which supporting multi win expr ptopN pushdown, it handled partial forbidden type unexpectly and will lead some case to push down the pTopN wrongly. plan before fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +------------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +------------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalPartitionTopN | | ----------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +------------------------------------------------------------------------------------+ plan after fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +----------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +----------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +----------------------------------------------------------------------------------+
github-actions bot
pushed a commit
that referenced
this pull request
Nov 27, 2024
…forbidden type (#44617) Related PR: #38393 Problem Summary: In the previous pr which supporting multi win expr ptopN pushdown, it handled partial forbidden type unexpectly and will lead some case to push down the pTopN wrongly. plan before fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +------------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +------------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalPartitionTopN | | ----------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +------------------------------------------------------------------------------------+ plan after fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +----------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +----------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +----------------------------------------------------------------------------------+
morrySnow
pushed a commit
that referenced
this pull request
Nov 27, 2024
…forbidden type (#44617) Related PR: #38393 Problem Summary: In the previous pr which supporting multi win expr ptopN pushdown, it handled partial forbidden type unexpectly and will lead some case to push down the pTopN wrongly. plan before fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +------------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +------------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalPartitionTopN | | ----------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +------------------------------------------------------------------------------------+ plan after fixing: explain shape plan select * from (select row_number() over(partition by c1, c2 order by c3) as rn, sum(c2) over(order by c2 range between unbounded preceding and unbounded following) as sw from push_down_multi_predicate_through_window_t) t where rn <= 1 and sw <= 1; +----------------------------------------------------------------------------------+ | Explain String(Nereids Planner) | +----------------------------------------------------------------------------------+ | PhysicalResultSink | | --PhysicalProject | | ----filter((rn <= 1) and (sw <= 1)) | | ------PhysicalWindow | | --------PhysicalQuickSort[MERGE_SORT] | | ----------PhysicalDistribute[DistributionSpecGather] | | ------------PhysicalQuickSort[LOCAL_SORT] | | --------------PhysicalWindow | | ----------------PhysicalQuickSort[LOCAL_SORT] | | ------------------PhysicalDistribute[DistributionSpecHash] | | --------------------PhysicalOlapScan[push_down_multi_predicate_through_window_t] | +----------------------------------------------------------------------------------+
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
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.
Support partitionTopn for multi window exprs.
If row_number exists, choose row_number with the minimal limit value; if not, choose others with the minimal limit value.
Example: