ulysses-you commented on code in PR #5617:
URL: https://github.com/apache/incubator-gluten/pull/5617#discussion_r1618117612
##########
shims/common/src/main/scala/org/apache/gluten/GlutenConfig.scala:
##########
@@ -208,7 +208,7 @@ class GlutenConfig(conf: SQLConf) extends Logging {
def queryFallbackThreshold: Int =
conf.getConf(COLUMNAR_QUERY_FALLBACK_THRESHOLD)
- def fallbackIgnoreRowToColumnar: Boolean =
conf.getConf(COLUMNAR_FALLBACK_IGNORE_ROW_TO_COLUMNAR)
+ def ignoreScanFallbackCost: Boolean =
conf.getConf(COLUMNAR_IGNORE_SCAN_FALLBACK_COST)
Review Comment:
> looks in this example, RowToCoumnar is not counted in the Join stage.
This is because we have `fallbackIgnoreRowToColumnar`.. if we drop
`fallbackIgnoreRowToColumnar` then this plan can not be transformed.
> User can increase the configured threshold if transformed plan is
preferred.
It's hard to let user tune config one by one. In general, we have a default
config that set transition cost to 1 and set fallbackIgnoreRowToColumnar to
true, so that all application prefer to use columnar and avoid performance
regression as far as possible.
--
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]
