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: 11706 vs ([lno] Canonical iv creation)



On Mar 12, 2004, at 08:42, Paolo Carlini wrote:


Hi again,

finally, I have been able to actually test the the new pass on the code
we care about. The current assembler, is simply *astonishing*.
E.g.:

double f(double a)
{
 return std::pow(a, 4);
}

On x86, -O2 -fomit-frame-pointer -ftree-loop-optimize -fscalar-evolutions
-funroll-loops:


00000000 <f(double)>:
  0:   dd 44 24 04             fldl   0x4(%esp)
  4:   d8 c8                   fmul   %st(0),%st
  6:   d8 c8                   fmul   %st(0),%st
  8:   c3                      ret

What is even more astonishing is this function: double f1() { return std::pow(3.0,7); } Before GCC would not optimize the function into just a return of a constant, now it does.


Thanks, Andrew Pinski


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