|Re: [eigen] sse asin implementation|
[ Thread Index |
| More lists.tuxfamily.org/eigen Archives
On Tue, 7 Apr 2009, Rohit Garg wrote:
Actually AFAIK, using more than 64 bits for doubles is _non_ compliant
with the IEEE spec.
I didn't check the standard itself, but I looked in Overton, Numerical
Computing with IEEE Floating Point Arithmetic, SIAM, and he says it's okay
to compute in extended precision (80 bits) and round to single or double
precision when storing the results in memory, as the Intel CPU does.
This is going rather off-topic though. Sorry.