http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56200
Alexander Monakov <amonakov at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |hjl.tools at gmail dot com,
| |ubizjak at gmail dot com
--- Comment #4 from Alexander Monakov <amonakov at gcc dot gnu.org> 2013-02-05
09:46:13 UTC ---
The need for the first alignment is clear: it aligns the loop to a 16-byte
boundary, and gcc does set that alignment at -O2. Uros, H.J., any idea why
separating the first conditional jump from the rest by additional alignment is
crucial for performance in this case? Is there anything that can be improved
in GCC here?
