andygrove commented on code in PR #1007:
URL: https://github.com/apache/datafusion-comet/pull/1007#discussion_r1793944208
##########
spark/src/test/resources/tpcds-plan-stability/approved-plans-v1_4-spark3_5/q16/simplified.txt:
##########
@@ -1,44 +1,59 @@
-WholeStageCodegen (2)
+WholeStageCodegen (7)
HashAggregate [sum,sum,count]
[sum(UnscaledValue(cs_ext_ship_cost)),sum(UnscaledValue(cs_net_profit)),count(cs_order_number),order
count ,total shipping cost ,total net profit ,sum,sum,count]
InputAdapter
Exchange #1
- WholeStageCodegen (1)
+ WholeStageCodegen (6)
HashAggregate [cs_order_number]
[sum(UnscaledValue(cs_ext_ship_cost)),sum(UnscaledValue(cs_net_profit)),count(cs_order_number),sum,sum,count,sum,sum,count]
HashAggregate [cs_order_number]
[sum(UnscaledValue(cs_ext_ship_cost)),sum(UnscaledValue(cs_net_profit)),sum,sum,sum,sum]
- ColumnarToRow
- InputAdapter
- CometHashAggregate
[cs_order_number,sum,sum,cs_ext_ship_cost,cs_net_profit]
- CometProject
[cs_order_number,cs_ext_ship_cost,cs_net_profit]
- CometBroadcastHashJoin
[cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cc_call_center_sk]
- CometProject
[cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
- CometBroadcastHashJoin
[cs_ship_addr_sk,cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,ca_address_sk]
- CometProject
[cs_ship_addr_sk,cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
- CometBroadcastHashJoin
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,d_date_sk]
- CometSortMergeJoin
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cr_order_number]
- CometProject
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
- CometSortMergeJoin
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cs_order_number,cs_warehouse_sk]
- CometSort
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
- CometExchange [cs_order_number] #2
- CometProject
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
- CometFilter
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cs_sold_date_sk]
- CometScan parquet
spark_catalog.default.catalog_sales
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cs_sold_date_sk]
- CometSort
[cs_warehouse_sk,cs_order_number]
- CometExchange [cs_order_number] #3
- CometProject
[cs_warehouse_sk,cs_order_number]
- CometScan parquet
spark_catalog.default.catalog_sales
[cs_warehouse_sk,cs_order_number,cs_sold_date_sk]
- CometSort [cr_order_number]
- CometExchange [cr_order_number] #4
- CometProject [cr_order_number]
- CometScan parquet
spark_catalog.default.catalog_returns [cr_order_number,cr_returned_date_sk]
- CometBroadcastExchange [d_date_sk] #5
- CometProject [d_date_sk]
- CometFilter [d_date_sk,d_date]
- CometScan parquet
spark_catalog.default.date_dim [d_date_sk,d_date]
- CometBroadcastExchange [ca_address_sk] #6
- CometProject [ca_address_sk]
- CometFilter [ca_address_sk,ca_state]
- CometScan parquet
spark_catalog.default.customer_address [ca_address_sk,ca_state]
- CometBroadcastExchange [cc_call_center_sk] #7
- CometProject [cc_call_center_sk]
- CometFilter [cc_call_center_sk,cc_county]
- CometScan parquet
spark_catalog.default.call_center [cc_call_center_sk,cc_county]
+ HashAggregate [cs_order_number,cs_ext_ship_cost,cs_net_profit]
[sum(UnscaledValue(cs_ext_ship_cost)),sum(UnscaledValue(cs_net_profit)),sum,sum,sum,sum]
+ Project [cs_order_number,cs_ext_ship_cost,cs_net_profit]
+ BroadcastHashJoin [cs_call_center_sk,cc_call_center_sk]
+ Project
[cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
+ BroadcastHashJoin [cs_ship_addr_sk,ca_address_sk]
+ Project
[cs_ship_addr_sk,cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
+ BroadcastHashJoin [cs_ship_date_sk,d_date_sk]
+ ShuffledHashJoin [cs_order_number,cr_order_number]
+ InputAdapter
+ WholeStageCodegen (1)
+ ColumnarToRow
+ InputAdapter
+ CometProject
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
+ CometHashJoin
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cs_order_number,cs_warehouse_sk]
+ CometExchange [cs_order_number] #2
+ CometProject
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit]
+ CometFilter
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cs_sold_date_sk]
+ CometScan parquet
spark_catalog.default.catalog_sales
[cs_ship_date_sk,cs_ship_addr_sk,cs_call_center_sk,cs_warehouse_sk,cs_order_number,cs_ext_ship_cost,cs_net_profit,cs_sold_date_sk]
+ CometExchange [cs_order_number] #3
+ CometProject
[cs_warehouse_sk,cs_order_number]
+ CometScan parquet
spark_catalog.default.catalog_sales
[cs_warehouse_sk,cs_order_number,cs_sold_date_sk]
+ InputAdapter
+ WholeStageCodegen (2)
+ ColumnarToRow
+ InputAdapter
+ CometExchange [cr_order_number] #4
+ CometProject [cr_order_number]
+ CometScan parquet
spark_catalog.default.catalog_returns [cr_order_number,cr_returned_date_sk]
+ InputAdapter
+ BroadcastExchange #5
+ WholeStageCodegen (3)
+ ColumnarToRow
+ InputAdapter
+ CometProject [d_date_sk]
+ CometFilter [d_date_sk,d_date]
+ CometScan parquet
spark_catalog.default.date_dim [d_date_sk,d_date]
+ InputAdapter
+ BroadcastExchange #6
+ WholeStageCodegen (4)
+ ColumnarToRow
+ InputAdapter
+ CometProject [ca_address_sk]
+ CometFilter [ca_address_sk,ca_state]
+ CometScan parquet
spark_catalog.default.customer_address [ca_address_sk,ca_state]
+ InputAdapter
+ BroadcastExchange #7
Review Comment:
This is a regression that I am looking into (falling back to Spark for
BroadcastHashJoin)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]