https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79128
Bug ID: 79128
Summary: export unsafe-math-optimization cpp defintion
Product: gcc
Version: 6.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: jtaylor.debian at googlemail dot com
Target Milestone: ---
GCC provides the __FAST_MATH__ when building with -ffast-math.
This allows c libraries to use other types of math functions that still comply
with this setting, e.g. vectorized variants.
A big issue is that -ffast-math is too aggressive for many programs, in
particular -ffinite-math is a big problem for many programs that could
otherwise live with more relaxed math.
It would be nice if GCC provides finer grained cpp definitions for libraries to
use.
E.g. -funsafe-math-optimizations or the subflags it enables are often
acceptable for numeric applications.
Having a definition available for libraries would allow for example to use
glibc libmvec without pulling -ffinite-math which you usually don't won't (note
until glibc 2.25 libmvec is compliant with -fno-finite-math, if gcc exports
more information it would be worthwhile to revert this change, making it much
more useful).
