https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86802
Jeffrey A. Law <law at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Resolution|--- |INVALID
Status|NEW |RESOLVED
--- Comment #5 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Thinking more about this, ISTM that each uarch will need to address this
problem or declare it as a non-issue. ie, the preferred sequence may
ultimately be different depending on the precise nature of the design. Some
high performance designs are expected to be immune to these problems (Ventana's
design for example).
It doesn't really seem useful to keep this ancient bug open.
