|
The included makefiles will make both floating point and fixed point versions of the testbed on a gnu based platform (such as linux or cygwin). For the DS version, the libraries will build automatically if you have the devkitpro environment set up (i.e. if the DEVKITPRO environment variables are set correctly). Currently there is no DS testbed.
Regarding the stability of the fixed point implementation, there are limitations in the fixed point implementation due to limitations of the number format used (generally variables are 32 bit between -32768 and +32765). The code does not handle all overflow and underflow cases but seems to work pretty well. In the testbed, only the latest example crashes due to math errors and I have a fix for those but haven't sent them out yet. That said, the more complex the example I'm sure you will find problems here and there.
I know of two programs that use the fixed point implementation on the DS that you can use to test if it is good enough for your application: pocketphysics and pinhockey. I have no information about the PSP.
|