harish a crit :

Quote:

> hi

> I was trying to do an DSP functions like FFT etc in hardware using VHDL.

> Since all my inputs and variables can only be bit vectors for them to be

> syntheziable I was wondering if a floating point math package which

> does FP mult, addn and sub is avaible on the net. What does everyone who

> does DSP on hardware do as the std package is sure not suffient to

> handle this?

> thanx for any help

> harish

How do you perform FFT on an integer DSP (no FPU)? You code it in a

fixed point way. Same with hardware implementations. Floating point

operations are usually avoidable in signal processing. There is some

work to be done about accuracy and dynamics but a lot of gates and

ns are saved.

Regards,

--

Renaud Pacalet, ENST / COMELEC, 46 rue Barrault 75634 Paris Cedex 13