https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116979

--- Comment #31 from Alexander Monakov <amonakov at gcc dot gnu.org> ---
(In reply to Alexander Monakov from comment #30)
> (I think with one extra fma it should be possible to compute sum of two
> products required here accurately, a*b+c*d -> fma(a, b, c*d)+fma(c, d,
> -c*d)) 

(this is Kahan's algorithm with 1.5ulp error, see the excellent answer by
Norbert Juffa at
https://stackoverflow.com/questions/63665010/accurate-floating-point-computation-of-the-sum-and-difference-of-two-products
)

Reply via email to