Single-Precision Software Floating Point?

[Ralf Corsepius]

On Thu, 2005-05-19 at 08:20 -0400, Mark Vander Voord wrote:
> I am using the MIPS CSB350 port, v 4.6.1

OK, mips ... not necessary a target I am familiar with ;)

> On 5/19/05, Ralf Corsepius <ralf.corsepius at rtems.org> wrote:
> > On Thu, 2005-05-19 at 08:08 -0400, Mark Vander Voord wrote:
> > > Is there any way to force single precision floats with the
> > > software-float library?  I see compiler flags for -msoft-float and
> > > -msingle-float, but it seems like specifying both still results in
> > > double precision.
> > Well, single precision is a dark chapter in GCC, all is highly target
> > specific, ...

... let me ask differently: What is this the problem you are trying to
solve? The mips has a plethora of fpu and 32/64 bit options which all
address different issues, so this isn't easy to answer.

AFAIU, you seem to be looking for an compiler option that implicitly
causes the compiler to use 4byte floats instead of 8byte floats.
>From what I see inside of the source code the mips-GCC doesn't have such
an option.

If your problem is memory, then -msoft-float and using "float" in your
application code instead of double could be what you are searching for.

Ralf