Regarding these packages... they are built using double-precision floating 
point arithmetic. I wonder if I should build the single precision version 
instead... What do you think?

-- 
Felipe Sateler