-msoft-float

David Boddie davidb at mcs.st-and.ac.uk
Tue Jun 5 09:44:40 PDT 2001


On 5 Jun 2001, Nick Burrett wrote:

> David Boddie <davidb at mcs.st-and.ac.uk> writes:
> 
> > On 5 Jun 2001, Nick Burrett wrote:
> > 
> > > David Boddie <davidb at mcs.st-and.ac.uk> writes:
> > > 
> > > > On 5 Jun 2001, Nick Burrett wrote:
> > > > 
> > > > > It is possible to do this, but it has never been fully tested.  However,
> > > > > I don't see the advantage of doing this, since code size will be bigger
> > > > > and the software floating point libraries are much slower than the FPE.
> > > > 
> > > > Can the latest version of gcc for RISC OS manage this?
> > > 
> > > It used to, but I turned it off. I decided it was pointless since UnixLib,
> > > the SharedClibrary and various RISC OS modules use FP instructions.
> > 
> > OK, so if an application spends a lot of time outside these then there
> > may be a gain.
> 
> No. But that's the problem, when you comment on a particular sentence rather
> than considering the e-mail as a whole, then replying to that. :-)

I'm not sure what you mean, but I was assuming that by using SCLib/UnixLib
less there is more potential for improvement whereas you presumably were
referring to the difficulty/disadvantages of building things with
soft-float support?

> > > One needs to fix a few files in GCC and it's Makefiles to get this working
> > > again.  It'd probably take an hour.
> > 
> > Can I do this, or does it require a certain level of experience with gcc?
> 
> You'll have to leave this bit to me. It requires knowledge about why
> files are stored in the directories they are in, and a bit of gcc knowledge.

Don't do anything yet. I'll take a look at the library first.

> > That may be more committment than I was hoping for. How big is the
> > library?
> 
> The library in question is gccsdk/gcc-2_95_2/gcc/config/fp-bit.c, which
> is 33512 bytes.  It does not support rounding modes, or exceptions, nor
> 96-bit floats.

OK. I have this.

> You'd probably need to hand-code the thing in assembler, and do some decent
> research into clever algorithms to reap the rewards.

Hmm.

> To get the latest CVS version, just do:
>   $ cvs update -A
> 
> At least then if I do make the changes for -msoft-float, you'll be able
> to get them.

Well, I'll familiarise myself with the library and possibly make some
changes, then perhaps we can discuss whether it is worthwhile making the
changes for -msoft-float. If that's OK with you?

-- 
David Boddie
Solar MHD Theory Group, St Andrews
http://www-solar.mcs.st-and.ac.uk/~davidb/index2.html




More information about the gcc mailing list