Re: weird optimization in sin+cos, x86 backend

[Vincent Lefevre <vincent+gcc at vinc17 dot org>]

On 2012-02-16 14:51:26 +0000, Andrew Haley wrote:
> Ah, I see.  So, we could bring gcc+glibc into compliance by not
> defining the rounding mode macros.

Yes, but... Directed rounding modes can also be used with the
arithmetic operations only, on which there are no problems at the
implementation level (well, there's still bug 34678: Optimization
generates incorrect code with -frounding-math option (#pragma STDC
FENV_ACCESS not implemented)). They could be defined only by some
option, and this could be related to -frounding-math. In particular,
the fact that directed rounding modes are affected by optimization
bugs may be another reason not to define these macros by default.

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)