On 7/14/06, Graham Murray <[EMAIL PROTECTED]> wrote:
"Richard Fish" <[EMAIL PROTECTED]> writes:
> Hrm, these are really not sane. -march is telling gcc to build C code
> that will only run on a p4, and then you have -mtune specifying to run
> on everything back to a pentium-II. I *think* -march takes precedence
> here....
Does it? I would have thought that having both -march and -mtune would
be valid, but normally would be other way round by having, for
example, '-march=i686 -mtune=pentium4' to build code which will run on
a pentium-II or later but optimised for running on a P4.
Right, this is the normal case. But in some sense, you could also
consider -march to have precedence here. If it ever needs to make a
choice between generating optimized code for a p4, or code that will
run on a P-II, gcc will choose the P-II compatible code.
But the fact is that I am not sure what -march=pentium4 -mtune=i686
would generate...
-Richard
--
gentoo-user@gentoo.org mailing list
