[clean-list] Handling of reals & boolean expressions?

RT Happe rthappe@mathematik.uni-freiburg.de
Fri, 20 Dec 2002 16:55:57 +0100 (CET)


On Fri, 20 Dec 2002, Richard A. O'Keefe wrote:

> If the "PC" field of the FPU control word is set to 53-bit mode,
> then the computed results should be the same AS IF they were
> converted to IEEE double, stored, and then loaded again.
>
> It might be worth offering a compiler option to set this mode.

To illustrate the practical value of less precise but predictable f.p.
arithmetic:  J.R.Shewchuk's triangulator and the underlying package for
arbitrary precision arithmetic rely on the absence of intermediate
extended precision results.  The C code works on Linux PCs thanks to the
gcc compiler switch forcing the desired behaviour.  (Or that is my
belief.)  Cf. http://www-2.cs.cmu.edu/~quake/robust.html

rthappe