This is the mail archive of the 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]


>>>>> "Jan" == Jan Hubicka <> writes:

    >>  Then, make for_each_rtx inline if we're compiling with GCC (as
    >> in stage2 and subsequent stages).  Then, if the compiler
    >> doesn't do what it should, improve the optimizers. :-)

    Jan> Avoiding penalties for data packed in the structures or
    Jan> globals can be tricky.

Yes, but that's what we compiler guys are supposed to be good at.
Where reasonable, we should try to program as we would like our users
to program.  If the only way to get good code out of GCC is with giant
macros-of-doom, then we haven't done our jobs properly.

    >>  I would much prefer this approach rather than the massive
    >> macro thing.

    Jan> Shame that we can't use other language here with more massive
    Jan> abstraction.

Yes, that is a shame.

I agree taht being able to include small expressions directly inline
seems convenient.  

Still, let's continue to use the function paradigm for now.  Frankly,
I think we can get a large amount of the speedup that way, and that
we'll get more over time as we improve the rest of the compiler.

Mark Mitchell         
CodeSourcery, LLC     

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