Karim Belabas on Wed, 25 Apr 2018 19:14:33 +0200

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

 Re: Comparison of multiple-precision floating-point software

```* paul zimmermann [2018-04-25 17:19]:
> > > https://members.loria.fr/PZimmermann/timings.html
> > >
> > > Pari is within a factor of 2 of MPFR for basic arithmetic, except for the
> > > square root, mul/sqr at 1000d, sqr/div at 10000d.
> > >
> > > Your feedback is welcome.
> >
> > Hi Paul,
> >
> > Thanks for the data and update !
> >
> > We already had this discussion a while back (in Medicis times...)
> > but I still believe you should enable the 3 precomputed constants
> > (log(2), Pi, Euler) in your timings-pari.c :
[...]
> feel free to try on your own. It is easy to get an account on the
> gcc67 machine.

Fair enough, that's what I ended up doing for the past 14 years or so
because we want to have timings for 32-bit and non-GMP pari anyway:

https://pari.math.u-bordeaux.fr/benchs/pari-2.9.0/timings-float.html

(PARI and MPFR only; outdated versions, sorry...).

Anyway, thanks again for sharing your benchmarks and results !

Cheers,

K.B.

P.S.
> I bet it will give at best a speedup of less than 1%, since for 10000
> digits we evaluate at least 127 times each function:

Possibly. If C is the time to compute a constant for a given accuracy
and F the time for a function call given the constant (for your specific input).
Then you measure C/127 + F, while I would measure F. So I agree with
your comment if C is of the order of F (or less, which it should be for
the tested functions).

--
Karim Belabas, IMB (UMR 5251)  Tel: (+33) (0)5 40 00 26 17
Universite de Bordeaux         Fax: (+33) (0)5 40 00 21 23
351, cours de la Liberation    http://www.math.u-bordeaux.fr/~kbelabas/
F-33405 Talence (France)       http://pari.math.u-bordeaux.fr/  [PARI/GP]
`

```