«

Hi,
I am looking for the fastest complex division algorithm available. It could be a white paper, a core that is being sold or any other information thatcould help me research different options. I will be using this in a spartan 3E with 4 complex multipliers available. Thanks for the help,
Amish

On 9/24/2012 11:52 AM, axr0284 wrote:
> I am looking for the fastest complex division algorithm
available. It could be a white paper, a core that is being sold or
any other information that could help me research different options.
I will be using this in a spartan 3E with 4 complex multipliers
available. Thanks for the help,

I don't have the answer you're looking for but it helps to avoid
hyperbolic adjectives like "fastest available". You should know your
requirements such that you can define "fast enough". Any faster is
wasted effort and expense.

I can imagine very wide combinatorial logic that will compute division
in one clock cycle. So can you. I doubt anyone could afford the
hardware to achieve that logic though.

JJS

Performance requirements should be identified in terms of latency and throughput. Very often, the lowest latency implementation does not have the highest throughput, and vise versa. A one-clock solution will almost certainly require a very slow clock.

Area requirements also need to be identified.

Andy

[jeppe]

If your not allowed to use automatic generated cores (like Logicore in Xilinx) then visit this page.

http://users-tima.imag.fr/cis/guyot/...ish/Divise.htm

Jeppe