sigsegv@rambler.ru wrote:
> I've noticed that during kernel compilation '-mno-fpu' flag is set. My 
> Ultra10 has 440MHz UltraSparc IIi cpu, which I'm sure has a built-in 
> FPU. Is it a good idea to tweak the Makefile to remove '-mno-fpu' so 
> that native hardware floating point instructions are used or kernel does 
> not use float and double datatypes?
> 
Looks like I've found the answer:

http://www.netbsd.org/Documentation/kernel/#why_msoft_float