Bill Allombert on Tue, 06 Jan 2004 14:35:23 +0100


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

PARI happy new year


Happy new year to all PARI lovers!

I have some news:

1) The 64bit bench was not updated after Karim last patch,
so make bench fail on 64bit machines. If the only problem
in the nfield diff is similar to (outside timing)

  ? getheap
! [200, 111766]
  ? print("Total time spent: ",gettime);
--- 1161,1167 ----
  ? getheap
! [200, 111754]

everything is fine.

2) I have worked on a public interface for polynomials modulo
small primes. The large patch is here:

<http://pari.math.u-bordeaux.fr/~bill/patch-flx-20040105>

In particular factormod should use less memory when the prime number
is small, on the other hand I have probably added several bugs.

The new function are prefixed with `Flx'. 'l' means a small prime
number and 'x' a 'small polynomial' stored as a t_VECSMALL.

This patch also remove some overload in the FpXQ functions that
should not be allowed.

3) Some discussions on the level 0 multi-precision kernel:

There were discussions with Guillaume Hanrot whether we could use GMP
longlong.h to implement the level0.h, but I never posted the conclusion.  

This is not possible since longlong.h implement a 2Nx2N->4N arithmetic
whereas PARI level0 need a NxN->2N arithmetic (where N=sizeof(long)),
so this is not suitable. On the other hand, Guillaume has written a
level0 kernel for ppc and ia64 so we have a level0 for all relevant
architectures currently.

Cheers,
Bill.