http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56741
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2013-03-26
Blocks| |53947
Summary|Why not to perform 128-bit |Epilogue loop not partly
|vector iteration when |vectorized
|vectorizing loop by 256-bit |
Ever Confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> 2013-03-26
14:22:55 UTC ---
Confirmed, but a very special case which will be difficult to handle. Note
that we need to know enough about the number of iterations of the epilogue
loop to make this even worthwhile (you'd peel a conditional execution of
one 128-bit vectorized iteration).
