Hello,

>     AFAIK, prior to NetBSD 1.3, the gcc included, added the option -softfloat
> when compiling with math support, thus making no use of the FPU.

>     Since 1.3 ALPHA, fpu support is complete (I think), so should I recompile
> all programs compiled with -softfloat, to gain true FP performance?

I'm not sure what you're talking about... There was no such restriction on 
pre-1.3 NetBSD/Amiga.

The only thing which has changed, is that NetBSD/Amiga has now (partial) 
support for FPU-less machines. FPU machines have been fully supported from 
day one.

The only use of -fsoft-float I know of, is when compiling the kernel. The
kernel itself doesn't issue FP operations (other than, to save, switch and
restore floating point context on behalf of user processes).

	-is