11706 vs ([lno] Canonical iv creation)

[Paolo Carlini]

Andrew Pinski wrote:

> 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.

Yes! And I'm seeing an excellent progress also for z^n, which is also very
important since we don't expect to have any time soon a builtin for integral
powers of complex numbers.

Paolo.