[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: pitfalls of verifying floating-point computations -- register spills
On 2007-07-19 23:02:02 +0000, Michel Hack wrote:
What are the possible differences in the three cases, in an
environment where registers use extended precision, as in x87?
It depends on the value of FLT_EVAL_METHOD. But IMHO, even if
FLT_EVAL_METHOD is 2, an implementation is still allowed to
produce results on 53-bit precision with a 64-bit long double
type since the C standard doesn't require correct rounding or
whatever.
See also 5.1.2.3#12, 6.3.1.5#2 and 6.3.1.8#2[52] (6.3.1.5#2 is
the normative part): after a cast or assignment, the value must
be converted to its semantic type (as opposed to the evaluation
type).
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.org/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/>
Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon)