Math trig intrinsics and compiler options

[gustav trede]

2009/7/15 Tim Bell <Tim.Bell at sun.com>

> gustav trede wrote:
>
> > If such test case exists, how do i get hold of it ?.
> > It would make testing of current compiler optimization status easier too.
>
> You could start with the open source Math and StrictMath tests in the jdk
> repository:
>
>  jdk/test/java/lang/Math/*
>  jdk/test/java/lang/StrictMath/*
>
> These tests might make interesting reading:
>
>  jdk/test/java/math/BigDecimal/*
>
>
> HTH-
>  Tim
>

I have already looked for Math cos and sin specific tests in the opendjk,
grep -RH --include '*'  'cos'  *
and i did not find it. i only find calls to those methods as part of other,
non math tests.
Several other trig methods do have some kind of tests in
test/java/lang/Math  ,
jdk/test/java/lang/StrictMath
Those completes successfully with this patch applied.

Azeem Jiva , a former hotspot developer that implemented some of the trig
intrinsics did propose this patch to me along with stating that to hes (old)
knowledge there should be no precision problems for math trig without the
rounding.

-
Thanks for all of yours time, trying to help and guide me.
 gustav trede
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-dev/attachments/20090716/06f78537/attachment.html