Simple question about MPFR vs. GMP

Richard B. Kreckel kreckel at ginac.de
Sat Apr 8 17:43:16 CEST 2006


Linas Vepstas wrote:

>By standarddizing to IEEE math, some of the "dumb" or "fixable"
>problems got fixed, but this doesn't mean that one can blithly 
>ignore sources of inaccuracy in your computations. IEEE rounding
>is not a magic cure-all.
>  
>

Obviously.  That reminds me of problem C14 at 
<http://www.cs.ru.nl/~milad/manydigits/problems.php>.

It's intersting that Paul and his team had to crank up the working 
precision in MPFR by approximately the same factor I had to crank it up 
in CLN (roughly 12000 decimal digits internal vs. 100 decimal digits 
precision goal).  If I understood correctly, the iRRAM team did not have 
to bother because that library did it correctly under the hood.  
Fascinating.

Regards
  -richy.

-- 
Richard B. Kreckel
<http://www.ginac.de/~kreckel/>



More information about the gmp-discuss mailing list