This is the mail archive of the
mailing list for the GCC project.
Re: Patch to add missing PPC instruction
- From: David Edelsohn <dje at watson dot ibm dot com>
- To: degger at fhm dot edu
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 17 Dec 2001 16:43:41 -0500
- Subject: Re: Patch to add missing PPC instruction
>>>>> degger writes:
degger> A candidate for -ffast-math? I originally meant it that way but forgot
degger> about it between the time I wrote the patch and sent it here
degger> (>4 months).
degger> For some applications (those who are using single precision arithmetics)
degger> it might actually be enough to such an unaccurate reciprocal but I
degger> definitely see your point when you say it's not acceptable to be used in
One part in 256 is too fast and loose for -ffast-math, IMHO. It
is meant as a graphics assist instruction.
If the programmer thinks this is useful, the programmer should use
the instruction directly; the compiler should not be substituting this
instruction except as part of a Newton-Raphson approximation to the