Re: Please listen to Ulrich here...
On 2013-08-09 09:47:56 +0200, Ulrich Kulisch wrote:
> I meant, of course, the EDP and in case of the evaluation a method by Rump
> and Boehm. I briefly discuss it in my book: Computer Arithmetic and
> Validity.
If you consider a software implementation, this is probably not the
best method. I've never seen the use of a dot product to evaluate
polynomials in real code.
If you consider a hardware implementation:
* A hardware EDP can make sense to evaluate polynomials, but only
with some particular representations and a particular algorithm,
and the time lost due to overhead is not clear.
* A hardware TwoSum (and possibly other similar transformations)
could be more useful.
* A hardware polynomial evaluation could also be considered and
would be faster.
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)