You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
currently we have completed supports to UDF/UDAF fallbacks. but UDAF is not yet supported. a HashAggregate with UDAF will fallback to spark.
in most cases, the bottlesnack of an aggregation is grouping. if we can only fallback the accumulating with UDAF, we may achieve more performance profits than fallback the whole operator.
Describe the solution you'd like
implement UDAF fallback. we have to consider how to reduce the number of JNI calls (one call per record is absolutely slow).
Describe alternatives you've considered
Additional context
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
currently we have completed supports to UDF/UDAF fallbacks. but UDAF is not yet supported. a HashAggregate with UDAF will fallback to spark.
in most cases, the bottlesnack of an aggregation is grouping. if we can only fallback the accumulating with UDAF, we may achieve more performance profits than fallback the whole operator.
Describe the solution you'd like
implement UDAF fallback. we have to consider how to reduce the number of JNI calls (one call per record is absolutely slow).
Describe alternatives you've considered
Additional context
The text was updated successfully, but these errors were encountered: