[Bussyman]

Hi,

It's right definition for float nan?

#if !defined FLOAT_NAN
#define FLOAT_NAN (Float:0xFFFFFFFF)
#endif

[Nero_3D]

Yes it is fine, all special values for 32 bit are

X111111111XXXXXXXXXXXXXXXXXXXXXX - quit NaN
X111111110XXXXXXXXXXXXXXXXXXXXXX - signaling NaN
01111111100000000000000000000000 - +infinity
11111111100000000000000000000000 - -infinity

the X's are up to you, either 0 or 1