[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gsl] Double vs Long Double
From: |
dave |
Subject: |
Re: [Help-gsl] Double vs Long Double |
Date: |
Fri, 5 Oct 2007 18:31:01 +0000 (GMT) |
On Fri, Oct 05, 2007 at 05:41:13PM +0100, Brian Gough wrote:
> At Thu, 4 Oct 2007 20:53:28 +0000 (GMT),
> dave wrote:
> > I tried setting GSL_IEEE_MODE='extended-precision' and the OpenBSD version
> > dumped core after printing an error message.
> >
> > gsl: fp-openbsd.c:41: ERROR: OpenBSD only supports default precision
> > rounding
> > Default GSL error handler invoked.
> > Abort trap (core dumped)
> >
> > This also happened with GSL_IEEE_MODE='double-precision'. It appears that
> > this
> > variable affects only rounding mode. It would be extremely handy if this
> > technique
> > could be used to switch between 80 bit temporary real (8087) math and 64
> > bit xmm math
> > operations.
>
> You can call the assembly instructions directly, details are in my
> book:
>
> http://www.network-theory.co.uk/docs/gccintro/gccintro_70.html
>
> --
> Brian Gough
With xmm registers present in an x86, fp can be computed either using the NPX
math unit
or else using the xmm registers. Would changing 'double' to 'DOUBLE' everwhere
in the GSL
source and then adding -DDOUBLE='long double' or -DDOUBLE='double' to the
CFLAGS work as a
means of generating either a library that used NPX temp real mode or a library
using XMM
double mode?
Thanks.
--