https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85186
Jeffrey A. Law <law at redhat dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |law at redhat dot com
--- Comment #6 from Jeffrey A. Law <law at redhat dot com> ---
A cost model would certainly be useful. I'm certainly open to ideas for
building that cost model.
Threading and vectorization do not play well together. We've gotten away with
throttling jump threading prior to vectorization, but it's never been a real
solution IMHO.
