[
https://issues.apache.org/jira/browse/HIVE-23133?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhenyu Zheng updated HIVE-23133:
--------------------------------
Attachment: HIVE-23133.6.patch
Status: Patch Available (was: Open)
> Numeric operations can have different result across hardware archs
> ------------------------------------------------------------------
>
> Key: HIVE-23133
> URL: https://issues.apache.org/jira/browse/HIVE-23133
> Project: Hive
> Issue Type: Sub-task
> Reporter: Zhenyu Zheng
> Assignee: Zhenyu Zheng
> Priority: Major
> Attachments: HIVE-23133.1.patch, HIVE-23133.2.patch,
> HIVE-23133.3.patch, HIVE-23133.4.patch, HIVE-23133.5.patch, HIVE-23133.6.patch
>
>
> Use java.lang.StrictMath rather than java.lang.Math in UDFs:
> UDFCos.java
> UDFExp.java
> UDFLn.java
> UDFLog.java
>
> Currently, we have set up an ARM CI to test out how Hive works on ARM
> platform:
> [https://builds.apache.org/view/H-L/view/Hive/job/Hive-linux-ARM-trunk/]
> Among the failures, we have observed that some numeric operations can have
> different result across hardware archs, such as:
> [https://builds.apache.org/view/H-L/view/Hive/job/Hive-linux-ARM-trunk/25/testReport/org.apache.hadoop.hive.cli/TestCliDriver/testCliDriver_vector_decimal_udf2_/]
> [https://builds.apache.org/view/H-L/view/Hive/job/Hive-linux-ARM-trunk/25/testReport/org.apache.hadoop.hive.cli/TestSparkCliDriver/testCliDriver_subquery_select_/]
> [https://builds.apache.org/view/H-L/view/Hive/job/Hive-linux-ARM-trunk/25/testReport/org.apache.hadoop.hive.cli/TestSparkCliDriver/testCliDriver_vectorized_math_funcs_/]
> we can see that the calculation results of log, exp, cos, toRadians etc is
> slitly different than the .out file results that we are
> comparing(they are tested and wrote on X86 machines), this is because of we
> use [Math
> Library|https://docs.oracle.com/javase/6/docs/api/java/lang/Math.html] for
> these kind of calculations.
> and according to the
> [illustration|https://docs.oracle.com/javase/6/docs/api/java/lang/Math.html]:
> _Unlike some of the numeric methods of class StrictMath, all implementations
> of the equivalent functions of class Math are not_
> _defined to return the bit-for-bit same results. This relaxation permits
> better-performing implementations where strict reproducibility_
> _is not required._
> _By default many of the Math methods simply call the equivalent method in
> StrictMath for their implementation._
> _Code generators are encouraged to use platform-specific native libraries or
> microprocessor instructions, where available,_
> _to provide higher-performance implementations of Math methods._
> so the result will have difference across hardware archs.
> On the other hand, JAVA provided another library
> [StrictMath|https://docs.oracle.com/javase/6/docs/api/java/lang/StrictMath.html]
> that will not have this kind of problem as according to its'
> [reference|https://docs.oracle.com/javase/6/docs/api/java/lang/StrictMath.html]:
> _To help ensure portability of Java programs, the definitions of some of the
> numeric functions in this package require that they produce_
> _the same results as certain published algorithms._
> So in order to fix the above mentioned problem, we have to consider switch to
> use StrictMath instead of Math.
>
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
