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: -mtune=generic for i386 backend


On Thu, Jan 19, 2006 at 06:13:19PM +0100, Jan Hubicka wrote:
> > > On Thu, Jan 19, 2006 at 12:15:47PM +0100, Jan Hubicka wrote:
> > > > > Jan Hubicka wrote:
> > > > > 
> > > > > > I wanted to express this by "is supposed to envolve in future versions
> > > > > > of GCC as new CPU models are introduced and other become obsolette.".
> > > > > > Since my english is poor, perhaps you can suggest better wording?
> > > > > 
> > > > > Actually, your English is very good, and a heck of a lot better than my
> > > > > Czech!  Your spelling needs a little work though; you might try M-x
> > > > > ispell-region in your friendly local emacs. :-) :-)
> > > > > 
> > > > > I've taken the liberty of reworking your paragraph.
> > > > > 
> > > > > "Produce code optimized for the most common IA32/AMD64/EM64T processors.
> > > > >    If you know the CPU on which your code will run, then you should use
> > > > > the corresponding @option{-mtune} option instead of
> > > > > @option{-mtune=generic}.  But, if you do not know exactly what CPU users
> > > > > of your application will have, then you should use this option.
> > > > > 
> > > > > As new processors are deployed in the marketplace, the behavior of this
> > > > > option will change.  Therefore, if you upgrade to a newer version of
> > > > > GCC, the code generated option will change to reflect the processors
> > > > > that were most common when that version of GCC was released.
> > > > > 
> > > > > There is no @option{-march=generic} option because @option{-march}
> > > > > indicates the instruction set the compiler can use, and there is no
> > > > > generic instruction set applicable to all processors.  In contrast,
> > > > > @option{-mtune} indicates the processor (or, in this case, collection of
> > > > > processors) for which the code is optimized."
> > > > > 
> > > > > Your patch is OK with that change, if you like the paragraphs I wrote. :-)
> > > > 
> > > > Thanks, I like your paragraphs ;)
> > > > Perhaps we can explicitely mention the set of CPUs we are tuning generic
> > > > for in current version of GCC, so people can answer the question whether
> > > > their CPU is included in our definition of "most common" and we also
> > > > fail to document "virtual" -march=x86-64 target, but this can be done as
> > > > followup.  I am going to finally test the patch with your paragraph and
> > > > H. J.'s fix to default CPU and commit it.
> > > > 
> > > 
> > > Jan, your last patch has some unrelated changes. Would you mind to
> > > post your final version before committing it?
> > 
> > I intend doing so.  I noticed one other minor latent problem in movqi
> > expansion where we didn't honnored -Os correctly.  This now cause
> > testsuite failure so I am re-testing patch with movqi fixed.  Will send
> > it once it looks OK.
> 
> This is what I finally comitted, hope I didn't missed anything...
> 

Thanks. I will rerun SPEC CPU 2K on my machines.


H.J.


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