hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Colin Patrick McCabe (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HADOOP-10778) Use NativeCrc32 only if it is faster
Date Mon, 07 Jul 2014 17:46:34 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-10778?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14053908#comment-14053908
] 

Colin Patrick McCabe commented on HADOOP-10778:
-----------------------------------------------

I like the idea of having a microbenchmark which can compare the different implementations.
 I'm not comfortable selecting the implementation to use at runtime with some microbenchmark,
because it means that behavior may be nondeterministic.  For example, if a GC hits when we're
doing the benchmark, or another process uses a bunch of CPu, we might choose the wrong implementation.
 As you guys know, most of our users on x86 just use CRC32C (not CRC32) with hardware acceleration,
and we're not going to beat that from Java or C.  Let's either get rid of the fallback mechanism
or add a configuration option to make it optional.

> Use NativeCrc32 only if it is faster
> ------------------------------------
>
>                 Key: HADOOP-10778
>                 URL: https://issues.apache.org/jira/browse/HADOOP-10778
>             Project: Hadoop Common
>          Issue Type: Improvement
>          Components: util
>            Reporter: Tsz Wo Nicholas Sze
>            Assignee: Tsz Wo Nicholas Sze
>         Attachments: c10778_20140702.patch
>
>
> From the benchmark post in [this comment|https://issues.apache.org/jira/browse/HDFS-6560?focusedCommentId=14044060&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14044060],
NativeCrc32 is slower than java.util.zip.CRC32 for Java 7 and above when bytesPerChecksum
> 512.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message