[MPFR] Seeking the fastest numerical library for quadruple precision in C/C++/Fortran

[Paul Zimmermann]

       Dear Michael,

there are several packages offering more than double precision:

- some machine vendors have implemented efficient quad precision in
  hardware (or at least in microcode). In particular Peter Markstein
  did implement a fast library on the Itanium. See the proceedings of
  the RNC5 conference (Lyon, 2003).
- Bayley et al. have designed a quad-double package, but as far as I understand
  they do not guarantee correct rounding even for addition and subtraction,
  and the last test I did, it was slower than mpfr for the same accuracy.
- there is recent work by Laurent and de Dinechin on a triple-double
  arithmetic package, which is quite efficient and portable. See for example
  <http://www.inria.fr/rrrt/rr-5702.html>.

Hope this helps,
Paul Zimmermann