This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFA, PATCH, PowerPC] Implements PowerPC e500mc core support


David Edelsohn wrote:

	Why is it not sufficient to configure GCC with e500 support if one
wants ISEL?  In other words, e500mc will work and produced code with
tuning, but ISEL only will be generated if configured for e500.

I don't think this should be an unconditionally static property of the compiler at configure time. We have users that want to use a single Power compiler to target a wide variety of Power CPUs, ranging from 603s to e500mc to G5s.


(In fact, I think we should be moving to more dynamism, not less. I would like to be able to target (say) ARM EABI and ARM GNU/Linux from the same compiler. However, that's a very big change.)

However, it seems reasonable to have (say) powerpc-* target all possible CPUs, while powerpc<something>-* targets only those CPUs indicated by <something>. For example, powerpcibm could target only IBM CPUs, or powerpcserver could target only server CPUs, or something. Or maybe we could do --with-cpus=603,440 to mean "only support the CPUs I've listed".

I don't have any preconceived notion about how to spell these things, or how'd they map to anything. I'm just trying to find a way that we can have both an uber-Power configuration in which all CPUs can be selected dynamically and a more-specific configuration that provides a smaller, faster compiler for a subset of CPUs.

--
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]