Karim BELABAS on Fri, 25 Oct 2002 16:01:54 +0200 (MEST)


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

Benchmarks


Hi,

  I found a benchmark including PARI/GP (gp-2.0.12, Lewis-Wester, see
[Benchmark] topic in my pari pages). Out of curiosity, I checked the current
development version against the bench.

Reflecting on my previous polredabs() post, it would be very nice to do this
more systematically: i.e extract some specialized topic, devise a
benchmark/testing suite to test relevant aspects, then apply it to PARI and
other (free or freely available) software.

For instance

* the PARI multiprecision kernel shows its age, and is slower than most
modern multiprecision libraries (GMP, CLN, ...). How slow ? In which range ?
Which routines ? Missing algorithms ? Before dumping the kernel for GMP, it
would be nice to be sure we don't have something interesting somewhere...

* polynomial arithmetic is slower than NTL. Factorization algorithms (over
Q, F_p, non-prime finite fields) are sometimes faster, sometimes slower.
Which ? When ? Why ?

* how do integer factorization routines compare with, e.g. GMP-ECM or PPMQS
(see the FAQ) ? primality provers ?

* LLL (integral/floating point) ?

* (complex) root finding ?

* more advanced stuff (e.g elliptic curves, alg. number theory, etc.)

The point is not to nominate world champions (any one project investing
enough development effort on a specific problem will eventually get the
crown), but to identify blatant problems, possibly spur development work,
and provide decent regression suites for PARI.

Of course, this will be done incrementally, and arbitrary areas can provide a
starting point.

Any takers ?

    Karim.

P.S: Maybe there are other "standard" benchs (available electronically...)
which would apply to PARI? Any idea?
-- 
Karim Belabas                    Tel: (+33) (0)1 69 15 57 48
Dép. de Mathematiques, Bat. 425  Fax: (+33) (0)1 69 15 60 19
Université Paris-Sud             Email: Karim.Belabas@math.u-psud.fr
F-91405 Orsay (France)           http://www.math.u-psud.fr/~belabas/
--
PARI/GP Home Page: http://www.parigp-home.de/