>Right in both cases .. maybe this should become and option
>--use-accelerator ... I think the floating point libraries will
>become faster with this option especially when we start
>using the accumulate functionality (i.e. 40 bit accumulator).
I haven't been active with SDCC recently, and it'll probably be
sometime in December when I actually do anything significant,
but I should let you know that I am planning to (eventually)
write a floating point library in hand-optimized mcs51 assembly.