# Direct3D vs. doubles

## Recommended Posts

Is it true that Direct3D disables doubles? I mean I'm almost sure it is so, but it is a problem to me: I'm creating a graphing tool with high precision (its input are latitude/longitude of a point in degrees). I found that setting precision back to 64 deals with the problem:
_control87 ( _PC_64, _MCW_PC );

The question is: can I do this only after Direct3D/device creation?

##### Share on other sites
xyzzy00    307
You might want to specify D3DCREATE_FPU_PRESERVE when you create your device. This tells D3D to save and restore the FPU state before and after each call into the API.

xyzzy

##### Share on other sites
Agony    3452
Check out D3DCREATE_FPU_PRESERVE, a flag used in the CreateDevice() call. Using this is probably a better method than manually changing it back yourself, or so I merely guess.

##### Share on other sites
DrGUI    402
.NET performance docs say that double is the fastest floating-point type because apparantly processors' FPUs have been 64-bit for a while. Is that true?

##### Share on other sites
I don't know about .NET docs, but as far as I know the FPUs have been 80-bit (like long double in Intel's compilers).

Anyway, thanx for help.