problem with lrintf()

jockyw2001 · Post by **jockyw2001** » Sun Sep 03, 2006 7:35 pm

I get an exception with lrintf(float f) if for example f=2264386731.980626
I tried to use llintf(double d), but that functions isn't present.

Is there perhaps a bug in the lrintf implementation?
(I saw this: http://www.cygwin.com/ml/cygwin/2005-09/msg00252.html)
What are actually the limits for f?
Does anyone know what alternative rounding function I can use?

Thx!

UPDATE: pls ignore my questions found the solution already, I must use lrint(double d)

dot_blank · Post by **dot_blank** » Fri Sep 08, 2006 6:34 pm

doubles on psp are a no no :P

jockyw2001 · Post by **jockyw2001** » Fri Sep 08, 2006 9:59 pm

Hmm, then please explain me why lrintf raises an exception and lrint works just fine?

chp · Post by **chp** » Fri Sep 08, 2006 10:11 pm

Probably because doubles are emulated by the compiler and won't trigger the same exception as the hardware float support does. Using software doubles cost a lot of performance however, so if this function is called often you will see performance degradation.

jockyw2001 · Post by **jockyw2001** » Fri Sep 08, 2006 10:20 pm

Okay, thanks for the insights guys