This is the mail archive of the gcc-help@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: Invariant is not moved out of loop


On 2017-07-25 18:04 +0200, Astor Piaz wrote:
> Hi Alexander,
> 
> Thank you, your answer is pretty clear and your example is much better.
> 
> Could I ask you though, why wouldn't this be allowed once the user has
> requested -fassociative-math ?

Alexander said this is not implemented, but it should be allowed. For
integer operands, this is allowed by ISO C/C++ standards.  For floating-
point operands, this should be allowed by -fassociative-math.

> Is there a fundamental problem to provide this optimization or it is
> just too difficult?

I don't know, but it seems no existing compilers has implemented it yet.

> Thanks a lot.
> 
> Best,
> --
> Astor
> 
> On Tue, Jul 25, 2017 at 5:11 PM, Alexander Monakov <amonakov@ispras.ru> wrote:
> > Hi,
> > 
> > I think what you're asking is not the "usual" loop invariant motion, but rather
> > applying distributive law to reductions. In your example you want the optimizer
> > to replace
> > 
> >   R = 0;
> >   for ( ... )
> >     R += X * C;
> > 
> > by
> > 
> >   R = 0;
> >   for ( ... )
> >     R += X;
> >   R *= C;
> > 
> > We don't do that even for integer operands, the following isn't optimized either:
> > (neither do Clang and ICC according to my experiments on gcc.godbolt.org)
> > 
> >   int f(int *a)
> >   {
> >     int r = 0;
> >     for (int i = 0; i < 1024; i++)
> >       r += a[i] * 5;
> >     return r;
> >   }
> > 
> > Alexander
-- 
Xi Ruoyao <ryxi@stu.xidian.edu.cn>
School of Aerospace Science and Technology, Xidian University


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