[Bug 1740] Change PowerPC multi-libs for e200 and e500 cores

[bugzilla-daemon at rtems.org]

https://www.rtems.org/bugzilla/show_bug.cgi?id=1740

Joel Sherrill <joel.sherrill at oarcorp.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |joel.sherrill at oarcorp.com

--- Comment #18 from Joel Sherrill <joel.sherrill at oarcorp.com> 2011-02-25 07:25:49 CST ---
(In reply to comment #17)
> (In reply to comment #14)
> 
> > Newlib needs IEEE 754 conformance - It doesn't support anything else.
> 
> Please expand.  Please explain what will happen when the non-spec shortcuts are
> in effect, and why it will not be useful.
> 
> Is it because carefully-written maths (the complex arc-cosine result shown
> earlier in the wrong quadrant) may be bogus?  If so, I think the heads-up that
> it isn't fully compliant without the software wrapper is adequate.
> 
> In the case of many real-time applications the injection of process noise is
> going to make the default behavior acceptable.  One will need to be careful
> with something like an ideal system model, and then injection of synthetic
> noise might be enough.  This is IMHO, I'd appreciate other opinions.
> 
> As Sebastian presented earlier, the behavior is outlined in Freescale's "Signal
> Processing
> Engine (SPE) Programming Environments Manual: 3.3.1.4", I have SPEPIM.pdf Rev.
> 0 01/2008

This is similar to cases in both the MIPS, 68040, and 68060.  If you don't have
an exception handler installed for extra precision and some operations, you
aren't IEEE754 compatible. 

I have emailed the authors of the exception package asking for a license
change.  Until we get the exception code, I guess this is caveat emptor.

-- 
Configure bugmail: https://www.rtems.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching all bug changes.