Re: Ada vs. C performance, was Possible to write compiler to

Toon Moene wrote:


> subroutine saxpy(n,sa,sx,incx,sy,incy)
> real sx(*),sy(*),sa
> integer i,incx,incy,ix,iy,n


<stuff deleted>


> The problem here, for the gcc backend, is that the arguments to the
> Fortran routine are implemented in g77 as "call-by-reference",
> i.e. pointers to their values in memory are passed. The C semantics
> doesn't promise anything about aliasing of pointers 1); therefore, the
> backend _currently_ assumes that the store into sy invalidates the
> values of sa, incx and incy in the loop.


I'm not sure if you are refering to the standard or the gcc
implementation, but just to make sure, the ANSI C standard "does
promise" some things about aliasing of pointers. For example, a
pointer to the real variable sy can not point to the integer variables
incx or incy (in an ANSI C compliant program). The gcc implementation
may invalidate incx and incy on a store to sy, but this is not
required by the ANSI C standard.


Best regards,
Christopher Glaeser cdg@nullstone.com
Nullstone Corporation http://www.nullstone.com
[True, but in this example, even the potential aliasing of variables
of the same type kills most optimizations. -John]


--

Return to the comp.compilers page.
Search the comp.compilers archives again.