jiaqizho commented on code in PR #1014:
URL: https://github.com/apache/cloudberry/pull/1014#discussion_r2018285377
##########
src/backend/utils/misc/guc_gp.c:
##########
@@ -2408,6 +2409,17 @@ struct config_bool ConfigureNamesBool_gp[] =
NULL, NULL, NULL
},
+ {
+ {"optimizer_force_split_window_function", PGC_USERSET,
QUERY_TUNING_METHOD,
Review Comment:
I think that is another problem in the ORCA cost model. The root case is:
**the estimated output rows in the cost of the window function were inaccurate
before this change**.
If the estimated number of output rows of the window function itself is
accurate, then the current cost model does not need to be adjusted. Becuase
current
The current global/local window function estimated output rows is 3670746 (a
large value). In fact, when the local window function enable, the cost of
motion has been reduced(because the input rows have been reduced). If the
current local window function estimated number of output rows is 1100 rows,
then ORCA will definitely choose the two-step window function path.
--
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: commits-unsubscr...@cloudberry.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cloudberry.apache.org
For additional commands, e-mail: commits-h...@cloudberry.apache.org
