Hi,

Without that, -m68040 does nothing. Together, you tell the compiler
> that it has an fpu unit to work with, and what kind. I speculate that
> even for integer arithmetic, it may make good use of the extra
> registers. We know that the compiler chooses to include floating point
> instructions for things that have nothing to do with arithmetic.
> That's why the --soft-float method needs the whole world to be built
> --soft-float.

I've tried all the optimization settings in the config and makefiles 
but, alas, it didn't improve performance at all. The new optimized 
kernel is just as slow with IPSec performance as the old kernel.

Too bad :-(

-Martin

----
This e-mail was arranged, composed and conducted
by M. van Hensbergen

[ PGP-key: http://home.wxs.nl/~mhens/mhens.key ]