> It would be great to use (chicken flonum) As it is now, fl+, fl*, etc. are defined in terms of (chicken flonum) procedures and foldr or foldl. It's possible that we could get a bit of a performance gain by internally just using (chicken flonum) procedures directly instead of relying on fl+ etc.
> and, where necessary, direct FFI calls and see if they are faster I agree about FFI procedures. I would very strongly expect them to be faster than the scheme procedures, but I'd rather not do it without an actual benchmark, which I haven't gotten around to yet. - Diego
