Andre wrote:
If the PPC has a 32x32=64bit multiply instruction,
It is two instructions (that are certainly pipelined, and could be exectued in parallel depending upon the core):
lower_part = low_multiply(x,y) upper_part = high_multiply(x,y)
This in in place of the single multiply when FPM_APPROX is used.
which might be a better solution than tweaking the PLL.
Nah...I always want to see how fast it can _really_ go :-).
Have fun :-)
(I should be doing other things....)
-- Dan