To: Sean Smith <S DOT D DOT Smith AT ecs DOT soton DOT ac DOT uk>
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu, jnc AT ecs DOT soton DOT ac DOT uk (John Carter),
        jap AT ecs DOT soton DOT ac DOT uk (Adrian Pickering),
        tmf AT ecs DOT soton DOT ac DOT uk (Tim Forcer)
Subject: Re: Portable FP emulation for DJGPP 
Date: Fri, 13 May 94 17:06:11 -0400
From: Mike Collison <collison AT sw DOT stratus DOT com>


> We have ported DJGPP to operate as a cross-compiler/environment
> for PC-hosted development of code for SPARC processors. Our environment
> is SPARC embedded system boards with parallel/serial IO, timers etc.
> for teaching embedded systems techniques to undergrads, and for
> developing embedded systems for internal/external projects.

> Our target boards have just the SPARC integer units on them, being
> simple embedded targets, but we would like to have the 'occasional'
> facility to use floating point via libraries.

> Compiling programs with GNU using '-mno-fpu' produces all the
> usual correct function calls to library routines that should
> emulate floating point. What we would like though are some
> portable C source code routines for implementing the floating
> point operations to IEEE specifications preferably.

> We could sit down and write them, but before reinventing the wheel,
> I just thought we ought to ask if anyone out there has come
> across a PD set of C or C++ sources that already do this?

Yeah, just look in 'floatlib.c' in the gcc source directory. It
could functions for emulating IEEE floating point arithmetic. Be
warned that it mostly only contains the single precision versions
of the emulation routines, but is a good starting point. I have used
the routines for a embedded processor and can tell you that they
work.

> Many thanks for looking!

Mike