[Bug tree-optimization/80198] [8/9/10 Regression] does not vectorize generic inplace integer operation

law at redhat dot com gcc-bugzilla@gcc.gnu.org
Mon Feb 24 23:30:00 GMT 2020


https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80198

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|8.4                         |11.0

--- Comment #18 from Jeffrey A. Law <law at redhat dot com> ---
So AFAICT opportunities to simplify as a result of conditional equivalences
that aren't used for propagation are pretty rare.  Call it a dozen across a
bootstrap.  While I only implemented binary operators, I'd be surprised if
adding other operators would help all that much. Maybe we'd get something from
INDIRECT_REF and MEM_REF.  Given we're well into stage4, it's hard to justify
adding this code right now.  I'll queue it for reevlauation in gcc-11 though --
the implementation isn't terribly complex, though I do worry about its
compile-time cost.


More information about the Gcc-bugs mailing list