hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Matt McCline (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-19269) Vectorization: Turn On by Default
Date Mon, 23 Apr 2018 04:26:00 GMT

    [ https://issues.apache.org/jira/browse/HIVE-19269?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16447552#comment-16447552
] 

Matt McCline commented on HIVE-19269:
-------------------------------------

Overwhelming.  Gave up at MiniSparkOnYarn.  Going to have to turn on vectorization on in
several Drivers at a time.
 * Unexpected Q output file differences:
 ** auto_join_without_localtask
 ** parquet_vectorization_0
 ** join0
 ** parallel_join0
 ** MiniTez: explainanalyze_1
 ** MiniTez: explainanalyze_4
 ** MiniTez: multi_count_distinct
 ** MiniTez: tez-tag
 ** MiniLlap: orc_ppd_basic
 ** MiniLlap: orc_ppd_schema_evol_3a
 ** MiniLlap: dynamic_semijoin_user_level
 ** MiniLlapLocal: orc_ppd_decimal
 ** 
 * Wrong results:
 ** Spark: bucketmapjoin9
 ** Spark: bucketmapjoin10
 ** Spark: semijoin
 ** Spark: skewjoin
 ** Spark: avro_decimal_native
 ** Spark: bucket_map_join_2
 ** Spark: bucket_map_join_1
 ** Spark: mapjoin_addjar
 ** Spark: mapjoin_decimal
 ** Spark: nullgroup
 ** MiniLlapLocal: offset_limit
 ** MiniLlapLocal: dynpart_sort_optimization
 ** MiniLlapLocal: lineage
 ** MiniLlapLocal: mapjoin46
 ** MiniLlapLocal: skiphf_aggr
 ** MiniLlapLocal: offset_limit_ppd_optimizer
 * Failures:
 ** Spark: bucket_map_join_spark1
 ** Spark: bucket_map_join_spark2
 ** Spark: bucketsortoptimize_insert_8
 ** Spark: bucketsortoptimize_insert_6
 ** Spark: bucketsortoptimize_insert_4
 ** Spark: decimal_join
 ** Spark: mapjoin1
 ** Spark: bucketmapjoin5
 ** Spark: bucketmapjoin8
 ** Spark: bucketmapjoin7
 ** Spark: bucketmapjoin11
 ** Spark: bucketmapjoin12
 ** Spark: bucketmapjoin13
 ** Spark: bucket_map_join_spark3
 ** MiniLlapLocal: tez_join_result_complex

MiniLlapLocal: retry_failure

> Vectorization: Turn On by Default
> ---------------------------------
>
>                 Key: HIVE-19269
>                 URL: https://issues.apache.org/jira/browse/HIVE-19269
>             Project: Hive
>          Issue Type: Bug
>          Components: Hive
>            Reporter: Matt McCline
>            Assignee: Matt McCline
>            Priority: Critical
>             Fix For: 3.0.0, 3.1.0
>
>         Attachments: HIVE-19269.01.patch
>
>
> Reflect that our most expected Hive deployment will be using vectorization and change
the default of hive.vectorized.execution.enabled to true.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message