[
https://issues.apache.org/jira/browse/CALCITE-6891?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alessandro Solimando updated CALCITE-6891:
------------------------------------------
Description:
physical plan not using this rule:
{code:java}
EnumerableIntersect(all=[false]): rowcount = 4.0, cumulative cost = {40.0 rows,
42.0 cpu, 0.0 io}, id = 37
EnumerableProject(DEPTNO=[$7]): rowcount = 14.0, cumulative cost = {28.0
rows, 29.0 cpu, 0.0 io}, id = 35
EnumerableTableScan(table=[[CATALOG, SALES, EMP]]): rowcount = 14.0,
cumulative cost = {14.0 rows, 15.0 cpu, 0.0 io}, id = 31
EnumerableProject(DEPTNO=[$0]): rowcount = 4.0, cumulative cost = {8.0 rows,
9.0 cpu, 0.0 io}, id = 36
EnumerableTableScan(table=[[CATALOG, SALES, DEPT]]): rowcount = 4.0,
cumulative cost = {4.0 rows, 5.0 cpu, 0.0 io}, id = 34
{code}
physical plan using this rule:
{code:java}
EnumerableIntersect(all=[false]): rowcount = 4.0, cumulative cost = {40.0 rows,
42.0 cpu, 0.0 io}, id = 39
EnumerableProject(DEPTNO=[$0]): rowcount = 4.0, cumulative cost = {8.0 rows,
9.0 cpu, 0.0 io}, id = 37
EnumerableTableScan(table=[[CATALOG, SALES, DEPT]]): rowcount = 4.0,
cumulative cost = {4.0 rows, 5.0 cpu, 0.0 io}, id = 32
EnumerableProject(DEPTNO=[$7]): rowcount = 14.0, cumulative cost = {28.0
rows, 29.0 cpu, 0.0 io}, id = 38
EnumerableTableScan(table=[[CATALOG, SALES, EMP]]): rowcount = 14.0,
cumulative cost = {14.0 rows, 15.0 cpu, 0.0 io}, id = 35
{code}
This rule put smaller inputs first. This helps reduce the size of intermediate
results.
We can see the diffferent of DAG, I used volcanol planner and topdown mode.
original
!image-2025-03-16-09-38-41-654.png|width=529,height=379!
used rule:
!image-2025-03-16-09-40-10-496.png|width=528,height=378!
was:
Do we need this rule?
phy plan no use this rule:
{code:java}
EnumerableIntersect(all=[false]): rowcount = 4.0, cumulative cost = {40.0 rows,
42.0 cpu, 0.0 io}, id = 37
EnumerableProject(DEPTNO=[$7]): rowcount = 14.0, cumulative cost = {28.0
rows, 29.0 cpu, 0.0 io}, id = 35
EnumerableTableScan(table=[[CATALOG, SALES, EMP]]): rowcount = 14.0,
cumulative cost = {14.0 rows, 15.0 cpu, 0.0 io}, id = 31
EnumerableProject(DEPTNO=[$0]): rowcount = 4.0, cumulative cost = {8.0 rows,
9.0 cpu, 0.0 io}, id = 36
EnumerableTableScan(table=[[CATALOG, SALES, DEPT]]): rowcount = 4.0,
cumulative cost = {4.0 rows, 5.0 cpu, 0.0 io}, id = 34
{code}
phy plan used this rule:
{code:java}
EnumerableIntersect(all=[false]): rowcount = 4.0, cumulative cost = {40.0 rows,
42.0 cpu, 0.0 io}, id = 39
EnumerableProject(DEPTNO=[$0]): rowcount = 4.0, cumulative cost = {8.0 rows,
9.0 cpu, 0.0 io}, id = 37
EnumerableTableScan(table=[[CATALOG, SALES, DEPT]]): rowcount = 4.0,
cumulative cost = {4.0 rows, 5.0 cpu, 0.0 io}, id = 32
EnumerableProject(DEPTNO=[$7]): rowcount = 14.0, cumulative cost = {28.0
rows, 29.0 cpu, 0.0 io}, id = 38
EnumerableTableScan(table=[[CATALOG, SALES, EMP]]): rowcount = 14.0,
cumulative cost = {14.0 rows, 15.0 cpu, 0.0 io}, id = 35
{code}
This rule put smaller inputs first. This helps reduce the size of intermediate
results.
We can see the diffferent of DAG, I used volcanol planner and topdown mode.
original
!image-2025-03-16-09-38-41-654.png|width=529,height=379!
used rule:
!image-2025-03-16-09-40-10-496.png|width=528,height=378!
> Implement IntersectReorderRule
> ------------------------------
>
> Key: CALCITE-6891
> URL: https://issues.apache.org/jira/browse/CALCITE-6891
> Project: Calcite
> Issue Type: Improvement
> Reporter: Zhen Chen
> Assignee: Zhen Chen
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.40.0
>
> Attachments: image-2025-03-16-09-37-47-916.png,
> image-2025-03-16-09-38-41-654.png, image-2025-03-16-09-40-10-496.png
>
>
> physical plan not using this rule:
> {code:java}
> EnumerableIntersect(all=[false]): rowcount = 4.0, cumulative cost = {40.0
> rows, 42.0 cpu, 0.0 io}, id = 37
> EnumerableProject(DEPTNO=[$7]): rowcount = 14.0, cumulative cost = {28.0
> rows, 29.0 cpu, 0.0 io}, id = 35
> EnumerableTableScan(table=[[CATALOG, SALES, EMP]]): rowcount = 14.0,
> cumulative cost = {14.0 rows, 15.0 cpu, 0.0 io}, id = 31
> EnumerableProject(DEPTNO=[$0]): rowcount = 4.0, cumulative cost = {8.0
> rows, 9.0 cpu, 0.0 io}, id = 36
> EnumerableTableScan(table=[[CATALOG, SALES, DEPT]]): rowcount = 4.0,
> cumulative cost = {4.0 rows, 5.0 cpu, 0.0 io}, id = 34
> {code}
> physical plan using this rule:
> {code:java}
> EnumerableIntersect(all=[false]): rowcount = 4.0, cumulative cost = {40.0
> rows, 42.0 cpu, 0.0 io}, id = 39
> EnumerableProject(DEPTNO=[$0]): rowcount = 4.0, cumulative cost = {8.0
> rows, 9.0 cpu, 0.0 io}, id = 37
> EnumerableTableScan(table=[[CATALOG, SALES, DEPT]]): rowcount = 4.0,
> cumulative cost = {4.0 rows, 5.0 cpu, 0.0 io}, id = 32
> EnumerableProject(DEPTNO=[$7]): rowcount = 14.0, cumulative cost = {28.0
> rows, 29.0 cpu, 0.0 io}, id = 38
> EnumerableTableScan(table=[[CATALOG, SALES, EMP]]): rowcount = 14.0,
> cumulative cost = {14.0 rows, 15.0 cpu, 0.0 io}, id = 35
> {code}
> This rule put smaller inputs first. This helps reduce the size of
> intermediate results.
> We can see the diffferent of DAG, I used volcanol planner and topdown mode.
> original
> !image-2025-03-16-09-38-41-654.png|width=529,height=379!
> used rule:
> !image-2025-03-16-09-40-10-496.png|width=528,height=378!
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
