/* mpn_toom_interpolate_7pts -- Interpolate for toom44, 53, 62.
Contributed to the GNU project by Niels Möller. Improvements by Marco Bodrato.
THE FUNCTION IN THIS FILE IS INTERNAL WITH A MUTABLE INTERFACE. IT IS ONLY SAFE TO REACH IT THROUGH DOCUMENTED INTERFACES. IN FACT, IT IS ALMOST GUARANTEED THAT IT WILL CHANGE OR DISAPPEAR IN A FUTURE GNU MP RELEASE.
Copyright 2006, 2007, 2009, 2014, 2015 Free Software Foundation, Inc.
This file is part of the GNU MP Library.
The GNU MP Library is free software; you can redistribute it and/or modify it under the terms of either:
* the GNU Lesser General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version.
or
* the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
or both in parallel, as here.
The GNU MP Library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received copies of the GNU General Public License and the GNU Lesser General Public License along with the GNU MP Library. If not,
see https://www.gnu.org/licenses/. */
/* For the various mpn_divexact_byN here, fall back to using either mpn_pi1_bdiv_q_1 or mpn_divexact_1. The former has less overhead and is many faster if it is native. For now, since mpn_divexact_1 is native on several platforms where mpn_pi1_bdiv_q_1 does not yet exist, do not use
mpn_pi1_bdiv_q_1 unconditionally. FIXME. */
/* For odd divisors, mpn_divexact_1 works fine with two's complement. */ #ifndef mpn_divexact_by3 #if HAVE_NATIVE_mpn_pi1_bdiv_q_1 #define mpn_divexact_by3(dst,src,size) mpn_pi1_bdiv_q_1(dst,src,size,3,BINVERT_3,0) #else #define mpn_divexact_by3(dst,src,size) mpn_divexact_1(dst,src,size,3) #endif #endif
/* Interpolation for toom4, using the evaluation points 0, infinity, 1, -1, 2, -2, 1/2. More precisely, we want to compute f(2^(GMP_NUMB_BITS * n)) for a polynomial f of degree 6, given the seven values
The result is 6*n + w6n limbs. At entry, w0 is stored at {rp, 2n }, w2 is stored at { rp + 2n, 2n+1 }, and w6 is stored at { rp + 6n, w6n }. The other values are 2n + 1 limbs each (with most significant limbs small). f(-1) and f(-1/2) may be negative, signs determined by the flag bits. Inputs are destroyed.
Note that most intermediate results are positive; the ones that may be negative are represented in two's complement. We must never shift right a value that may be negative, since that would invalidate the sign bit. On the other hand, divexact by odd numbers work fine with two's complement.
*/
/* These bounds are valid for the 4x4 polynomial product of toom44,
* and they are conservative for toom53 and toom62. */
ASSERT (w1[2*n] < 2);
ASSERT (w2[2*n] < 3);
ASSERT (w3[2*n] < 4);
ASSERT (w4[2*n] < 3);
ASSERT (w5[2*n] < 2);
/* Addition chain. Note carries and the 2n'th limbs that need to be * added in. * * Special care is needed for w2[2n] and the corresponding carry, * since the "simple" way of adding it all together would overwrite * the limb at wp[2*n] and rp[4*n] (same location) with the sum of * the high half of w3 and the low half of w4. * * 7 6 5 4 3 2 1 0 * | | | | | | | | | * ||w3 (2n+1)| * ||w4 (2n+1)| * ||w5 (2n+1)| ||w1 (2n+1)| * + | w6 (w6n)| ||w2 (2n+1)| w0 (2n) | (share storage with r) * ----------------------------------------------- * r | | | | | | | | | * c7 c6 c5 c4 c3 Carries to propagate
*/
cy = mpn_add_n (rp + n, rp + n, w1, m);
MPN_INCR_U (w2 + n + 1, n , cy);
cy = mpn_add_n (rp + 3*n, rp + 3*n, w3, n);
MPN_INCR_U (w3 + n, n + 1, w2[2*n] + cy);
cy = mpn_add_n (rp + 4*n, w3 + n, w4, n);
MPN_INCR_U (w4 + n, n + 1, w3[2*n] + cy);
cy = mpn_add_n (rp + 5*n, w4 + n, w5, n);
MPN_INCR_U (w5 + n, n + 1, w4[2*n] + cy); if (w6n > n + 1)
{
cy = mpn_add_n (rp + 6*n, rp + 6*n, w5 + n, n + 1);
MPN_INCR_U (rp + 7*n + 1, w6n - n - 1, cy);
} else
{
ASSERT_NOCARRY (mpn_add_n (rp + 6*n, rp + 6*n, w5 + n, w6n)); #if WANT_ASSERT
{
mp_size_t i; for (i = w6n; i <= n; i++)
ASSERT (w5[n + i] == 0);
} #endif
}
}
Messung V0.5
¤ Dauer der Verarbeitung: 0.11 Sekunden
(vorverarbeitet)
¤
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.