home *** CD-ROM | disk | FTP | other *** search
-
- Using a math-coprocessor could be a good idea.
- Of course it does handle float numbers, but it's _VERY_
- slow: for an addition, you have to wait about 100 cycles...
- We should use a math-coprocessor if we can't do without it,
- or if a high precision is needed. It depends of what is needed.
- Fixed numbers (emulation done using 32 bits integers - or even
- 16 bits - ) work quiet good.
-
- Fabrice.
-
-
-
- ################################################################
- # _______ _______ | Fabrice GIRARDOT | #
- # / / /--('')--\ \ \ | C,ASM MC680x0,DSP56001 | #
- # / / / / \/ \ \ \ \ | Real-time processing | #
- # ///// \____/ \\\\\ #
- # / \ mailto:Fabrice.Girardot@imag.fr #
- #F030! /T\ /T\ http://www-cami.imag.fr/~girardot/ #
- ################################################################
-
-