On Mon, 13 Mar 2006, Jay Austad wrote:

> Pari looks interesting.  I'm reading the docs, and it doesn't look
> like it will handle the size of the numbers I'm dealing with
> (2^4000000000 and larger).  I've used some approximation techniques
> to get my answer, however, I've tried 3 of them, and they all give me
> numbers that are way different.  I need an exact answer.  Eventually,
> I need a way to do this calculation fast too.  Does anyone here have
> a really good understanding of math and experience with FPGA's?

I comment that simply exactly *representing* a number of the form 
2^4000000000 takes like 476.8 megabytes.  Unless you're on a 64-bit system 
with boatloads of memory, you're going to be hitting memory problems big 
time- simply holding more then 2 or 3 numbers of that size in memory is 
going to be a trick.

I'd probably be looking at hand coding a solution using GMP and C in your 
place.  You're definately in FFT-multiply area.  Heck, simply *adding* two 
numbers is going to take a signifigant fraction of a second on most 
machines.

Brian