Abhishek,

>I need to implement 200 256-point FFT's for >real valued data per second.
So you need to do one 256-point transform in 5 msec.

>Is that too much for an 8-bit 8051 ??
Yes, it is too much. I do MAC in 20 nanoseconds on TI TMS5402 DSP 100 MHz, so for your case I need in 256*20=5120 nsec=5.12usec for one frequency point. it takes 256*5.12=1.3msec approximately for whole transformation.

If you don't want to do something else you can reduce DSP clock. Of course my algorithm is not much optimised but you can take this values into consideration.

Cheers,
George