This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: State of m68k float emulation in lb1sf68.asm (muldf, divdf)
- From: <tm_gccmail at kloo dot net>
- To: Jim Wilson <wilson at specifixinc dot com>
- Cc: Pavel Pisa <pisa at cmp dot felk dot cvut dot cz>, gcc at gcc dot gnu dot org
- Date: Mon, 17 Nov 2003 11:16:37 -0800 (PST)
- Subject: Re: State of m68k float emulation in lb1sf68.asm (muldf, divdf)
On Sun, 16 Nov 2003, Jim Wilson wrote:
> Pavel Pisa wrote:
> > I have looked into lb1sf68.asm and I have found that implementation
> > is very ineffective. It can be many times improved at least for CPU32
> > targets I expect 6-20 times improvement.
>
> FP emulation is a complicated business. If you want to get into this
> area, you should do some reading on computer arithmetic if you already
> haven't.
>
> If lb1sf68 is that bad, then why not just use fp-bit.c like most other
> targets do? lb1sf68 has some fp exception stuff that fp-bit.c doesn't
> have, but if you don't need it, you are probably better off with fp-bit.c.
Uh, have you looked at fp-bit.c? It's glacially slow.
> If you really want something better than fp-bit.c, then I would suggest
> looking into GMP. You might need to do a m68k port of it. I don't see
> any m68k code in the longlong.h file. This helps with the portability,
> as you would define different longlong.h macros for different members of
> the m68k family.
GMP isn't IEEE754 compatible, iirc. It's an arbitrary precision bignum
library, and therefore unsuitable for as an fp-bit.c replacement.
Toshi