Mailing-List: contact issues-help@spark.apache.org; run by ezmlm
Precedence: bulk
Date: Tue, 19 Jul 2016 13:14:20 +0000 (UTC)
From: =?utf-8?Q?Maciej_Bry=C5=84ski_=28JIRA=29?= <jira@apache.org>
To: issues@spark.apache.org
Message-ID: <JIRA.12985657.1467271650000.71833.1468934060461@Atlassian.JIRA>
In-Reply-To: <JIRA.12985657.1467271650000@Atlassian.JIRA>
References: <JIRA.12985657.1467271650000@Atlassian.JIRA> <JIRA.12985657.1467271650715@arcas>
Subject: [jira] [Comment Edited] (SPARK-16321) Pyspark 2.0 performance drop
 vs pyspark 1.6
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
archived-at: Tue, 19 Jul 2016 13:14:22 -0000


    [ https://issues.apache.org/jira/browse/SPARK-16321?page=3Dcom.atlassia=
n.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D153=
83889#comment-15383889 ]=20

Maciej Bry=C5=84ski edited comment on SPARK-16321 at 7/19/16 1:13 PM:
-----------------------------------------------------------------

Thanks.
It works for PushedFilter ([GreaterThan(id,9000000)])
But it's not the reason.=20
Time of execution is the same as with IsNotNull predicate.

I have no more ideas.


was (Author: maver1ck):
Thanks.
It works for PushedFilter ([GreaterThan(id,9000000)])
But it's not the reason.=20
Time of execution is the same as with IsNotNull predicate.

> Pyspark 2.0 performance drop vs pyspark 1.6
> -------------------------------------------
>
>                 Key: SPARK-16321
>                 URL: https://issues.apache.org/jira/browse/SPARK-16321
>             Project: Spark
>          Issue Type: Bug
>          Components: PySpark
>    Affects Versions: 2.0.0
>            Reporter: Maciej Bry=C5=84ski
>         Attachments: Spark16.nps, Spark2.nps, spark16._trace.png, spark16=
_query.nps, spark2_nofilterpushdown.nps, spark2_query.nps, spark2_trace.png=
, visualvm_spark16.png, visualvm_spark2.png, visualvm_spark2_G1GC.png
>
>
> I did some test on parquet file with many nested columns (about 30G in
> 400 partitions) and Spark 2.0 is 2x slower.
> {code}
> df =3D sqlctx.read.parquet(path)
> df.where('id > some_id').rdd.flatMap(lambda r: [r.id] if not r.id %100000=
 else []).collect()
> {code}
> Spark 1.6 -> 2.3 min
> Spark 2.0 -> 4.6 min (2x slower)
> I used BasicProfiler for this task and cumulative time was:
> Spark 1.6 - 4300 sec
> Spark 2.0 - 5800 sec
> Should I expect such a drop in performance ?
> I don't know how to prepare sample data to show the problem.
> Any ideas ? Or public data with many nested columns ?


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org