C language hacking

[dik at turing.UUCP]

>The reason why this matters should be pretty obvious to anyone who
>does graphics for a living.
In that case it is simple.  You do not need large parts of the fancy stuff
executed by standard sine and cosine routines; e.g. you will not need a
16 digit precision (as delivered by the standard Vax Unix sine and cosine).
Moreover, you may make quite sure that the argument to these functions
is within a range like [-pi,pi], so range reduction is also much simpler
to do.

It is in general a pain that all standard functions in C deliver a result
in double precision (what?, oh yes, the attempt is made) although you need
it in single precision, which results in a waste of time.
-- 
dik t. winter
centrum voor wiskunde en informatica
postbus 4079
1009 AB  amsterdam
nederland
+31 20 592 4102 (polish your dutch)
UUCP: decvax!mcvax!turing!dik