On 6/21/11 9:38 AM, David Martindale wrote:
16 bits in is 64K *entries* of 16 bits each, a total of 1 megabit of
ROM. The usual 90/180 degree folding could reduce that to 256 kbit.
Dave
On Tue, Jun 21, 2011 at 7:49 AM, Jim Lux wrote:
So then, it's a trade between a big ROM LUT or a bunc
16 bits in is 64K *entries* of 16 bits each, a total of 1 megabit of
ROM. The usual 90/180 degree folding could reduce that to 256 kbit.
Dave
On Tue, Jun 21, 2011 at 7:49 AM, Jim Lux wrote:
> So then, it's a trade between a big ROM LUT or a bunch o'gates for CORDIC.
> And for big N the bun
> So then, it's a trade between a big ROM LUT or a bunch o'gates for
> CORDIC. And for big N the bunch o'gates is probably going to be easier:
> 16 bit phase and wanting 16 bits out would be a 64kbit ROM (assuming you
> didn't do the usual thing of only needing 1/4 cycle)..
There is also th
Ohh..
It just came to me..
You're not using CORDIC as a replacement for both the phase accumulator
and cos LUT, but JUST instead of the LUT, so you ARE doing the "give me
cos(theta)" on every sample.
So then, it's a trade between a big ROM LUT or a bunch o'gates for
CORDIC. And for big N th