commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Gilles (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (MATH-740) Some "FastMath" functions are slow
Date Tue, 04 Apr 2017 16:04:41 GMT

    [ https://issues.apache.org/jira/browse/MATH-740?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15955303#comment-15955303
] 

Gilles commented on MATH-740:
-----------------------------

bq. Math.ulp(), Math.IEEEremainder(), Math.getExponent()

Those 3 are implemented in {{FastMath}}.

bq. Math.powerOfTwoD()

What's this?

bq. Also GWT 2.8 supports Math.doubleToLongBits() (probably very slow) but not Math.doubleToRawLongBits()

I'm lost here; there are no such methods in the JDK's {{Math}} class.

Concerning the subject of this issue (i.e. performance), is someone willing to start a discussion
on the "dev" list?

Practically, we should first set up JMH code to provide performance numbers, since discrepancies
among some "home-made" micro-benchmarking codes were never explained (nor did they elicit
much interest, as strange as it is, given the heated debate we had about FastMath-related
code...).


> Some "FastMath" functions are slow
> ----------------------------------
>
>                 Key: MATH-740
>                 URL: https://issues.apache.org/jira/browse/MATH-740
>             Project: Commons Math
>          Issue Type: Wish
>            Reporter: Gilles
>            Priority: Minor
>             Fix For: 4.0
>
>         Attachments: jafamatests.zip
>
>
> From the two benchmarks we currently have in "FastMathTestPerfomance", we have that the
following functions are much slower in "FastMath" than in either "Math" or "StrictMath" (the
performance *loss*, for each of the benchmarks, is given in parentheses):
> * log10 (46%, 36%)
> * log1p (68%, 112%)
> * tan (11%, 61%)
> * atan (26%, 125%)
> * atan2 (44%, 40%)



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message