This is the mail archive of the
mailing list for the GCC project.
Re: -mfpmath=sse -mno-80387 troubles
On Thu, 27 Jan 2005 17:25:01 +0100, Uros Bizjak <email@example.com> wrote:
> Just a note - do not use -mno-80387 if you are calling external math
> functions. This option now implies -mno-fp-ret-in-387 and it will
> introduce an ABI mismatch. The goal is that -mfpmath=sse should be
> enough to prevent unnecessary x87 code.
That code has been splitted to take care of that (and the fact that i
can't use -ffast-math because i want to handle NaNs), so there's only
a couple of (self contained) units with those switches. And i was
using it only because of such code gen problems. I'd be more than
happy to get rid of it :)
> There is nothing wrong, if fldl is replaced with fldz or fld1. The
> performance problems will arise in case when memory location is used in
> subsequent SSE computations. In this case, it would be better if zero is
> "generated" in SSE register and stored to memory from SSE reg.
Indeed, i've just reported cases where it was obviously counterproductive.
BTW the return-in-xmm0 attribute sounds quite handy, could also solve
in a more elegant fashion that
I'll keep my finger crossed for pragma to control loop unrolling and stuff :)