[BACK]Return to bz_divrem_n.c CVS log [TXT][DIR] Up to [local] / OpenXM_contrib / gmp / mpn / generic

Annotation of OpenXM_contrib/gmp/mpn/generic/bz_divrem_n.c, Revision 1.1.1.1

1.1       maekawa     1: /* mpn_bz_divrem_n and auxilliary routines.
                      2:
                      3:    THE FUNCTIONS IN THIS FILE ARE INTERNAL FUNCTIONS WITH MUTABLE
                      4:    INTERFACES.  IT IS ONLY SAFE TO REACH THEM THROUGH DOCUMENTED INTERFACES.
                      5:    IN FACT, IT IS ALMOST GUARANTEED THAT THEY'LL CHANGE OR DISAPPEAR IN A
                      6:    FUTURE GNU MP RELEASE.
                      7:
                      8:
                      9: Copyright (C) 2000 Free Software Foundation, Inc.
                     10: Contributed by Paul Zimmermann.
                     11:
                     12: This file is part of the GNU MP Library.
                     13:
                     14: The GNU MP Library is free software; you can redistribute it and/or modify
                     15: it under the terms of the GNU Lesser General Public License as published by
                     16: the Free Software Foundation; either version 2.1 of the License, or (at your
                     17: option) any later version.
                     18:
                     19: The GNU MP Library is distributed in the hope that it will be useful, but
                     20: WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
                     21: or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
                     22: License for more details.
                     23:
                     24: You should have received a copy of the GNU Lesser General Public License
                     25: along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
                     26: the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
                     27: MA 02111-1307, USA. */
                     28:
                     29: #include "gmp.h"
                     30: #include "gmp-impl.h"
                     31:
                     32: /*
                     33: [1] Fast Recursive Division, by Christoph Burnikel and Joachim Ziegler,
                     34:     Technical report MPI-I-98-1-022, october 1998.
                     35:     http://www.mpi-sb.mpg.de/~ziegler/TechRep.ps.gz
                     36: */
                     37:
                     38: static mp_limb_t mpn_bz_div_3_halves_by_2 _PROTO ((mp_ptr, mp_ptr, mp_srcptr,
                     39:                                                    mp_size_t, mp_ptr));
                     40:
                     41: static mp_limb_t mpn_bz_divrem_aux _PROTO ((mp_ptr, mp_ptr, mp_srcptr,
                     42:                                             mp_size_t, mp_ptr));
                     43:
                     44: /* mpn_bz_divrem_n(n) calls 2*mul(n/2)+2*div(n/2), thus to be faster than
                     45:    div(n) = 4*div(n/2), we need mul(n/2) to be faster than the classic way,
                     46:    i.e. n/2 >= KARATSUBA_MUL_THRESHOLD */
                     47: #ifndef BZ_THRESHOLD
                     48: #define BZ_THRESHOLD (7 * KARATSUBA_MUL_THRESHOLD)
                     49: #endif
                     50:
                     51: #if 0
                     52: static
                     53: unused_mpn_divrem (qp, qxn, np, nn, dp, dn)
                     54:      mp_ptr qp;
                     55:      mp_size_t qxn;
                     56:      mp_ptr np;
                     57:      mp_size_t nn;
                     58:      mp_srcptr dp;
                     59:      mp_size_t dn;
                     60: {
                     61:   /* This might be useful: */
                     62:   if (qxn != 0)
                     63:     {
                     64:       mp_limb_t c;
                     65:       mp_ptr tp = alloca ((nn + qxn) * BYTES_PER_MP_LIMB);
                     66:       MPN_COPY (tp + qxn - nn, np, nn);
                     67:       MPN_ZERO (tp, qxn);
                     68:       c = mpn_divrem (qp, 0L, tp, nn + qxn, dp, dn);
                     69:       /* Maybe copy proper part of tp to np?  Documentation is unclear about
                     70:         the returned np value when qxn != 0 */
                     71:       return c;
                     72:     }
                     73: }
                     74: #endif
                     75:
                     76: /* mpn_bz_divrem_n - Implements algorithm of page 8 in [1]: divides (np,2n)
                     77:    by (dp,n) and puts the quotient in (qp,n), the remainder in (np,n).
                     78:    Returns most significant limb of the quotient, which is 0 or 1.
                     79:    Requires that the most significant bit of the divisor is set.  */
                     80:
                     81: mp_limb_t
                     82: #if __STDC__
                     83: mpn_bz_divrem_n (mp_ptr qp, mp_ptr np, mp_srcptr dp, mp_size_t n)
                     84: #else
                     85: mpn_bz_divrem_n (qp, np, dp, n)
                     86:      mp_ptr qp;
                     87:      mp_ptr np;
                     88:      mp_srcptr dp;
                     89:      mp_size_t n;
                     90: #endif
                     91: {
                     92:   mp_limb_t qhl = 0;
                     93:   if (mpn_cmp (np + n, dp, n) >= 0)
                     94:     {
                     95:       qhl = 1;
                     96:       mpn_sub_n (np + n, np + n, dp, n);
                     97:       abort ();
                     98:     }
                     99:   if (n % 2 != 0)
                    100:     {
                    101:       /* divide (2n - 2) most significant limbs from np by those (n - 1) from dp */
                    102:       if (n < BZ_THRESHOLD)
                    103:        qhl += mpn_sb_divrem_mn (qp + 1, np + 2, 2 * (n - 1), dp + 1, n - 1);
                    104:       else
                    105:        qhl += mpn_bz_divrem_n (qp + 1, np + 2, dp + 1, n - 1);
                    106:       /* now (qp + 1, n - 1) contains the quotient of (np + 2, 2n - 2) by
                    107:         (dp + 1, n - 1) and (np + 2, n - 1) contains the remainder */
                    108:       if (mpn_sub_1 (np + n, np + n, 1,
                    109:                     mpn_submul_1 (np + 1, qp + 1, n - 1, dp[0])))
                    110:        {
                    111:          /* quotient too large */
                    112:          qhl -= mpn_sub_1 (qp + 1, qp + 1, n - 1, 1L);
                    113:          if (mpn_add_n (np + 1, np + 1, dp, n) == 0)
                    114:            { /* still too large */
                    115:              qhl -= mpn_sub_1 (qp + 1, qp + 1, n - 1, 1L);
                    116:              mpn_add_n (np + 1, np + 1, dp, n); /* always carry here */
                    117:            }
                    118:        }
                    119:       /* now divide (np, n + 1) by (dp, n) */
                    120:       qhl += mpn_add_1 (qp + 1, qp + 1, n - 1,
                    121:                        mpn_sb_divrem_mn (qp, np, n + 1, dp, n));
                    122:     }
                    123:   else
                    124:     {
                    125:       mp_ptr tmp;
                    126:       mp_size_t n2 = n/2;
                    127:       TMP_DECL (marker);
                    128:       TMP_MARK (marker);
                    129:       tmp = (mp_ptr) TMP_ALLOC (n * BYTES_PER_MP_LIMB);
                    130:       qhl = mpn_bz_div_3_halves_by_2 (qp + n2, np + n2, dp, n2, tmp);
                    131:       qhl += mpn_add_1 (qp + n2, qp + n2, n2,
                    132:                       mpn_bz_div_3_halves_by_2 (qp, np, dp, n2, tmp));
                    133:       TMP_FREE (marker);
                    134:     }
                    135:   return qhl;
                    136: }
                    137:
                    138: /* Like mpn_bz_divrem_n, but without memory allocation.  Also
                    139:    assumes mpn_cmp (np + n, dp, n) < 0 */
                    140:
                    141: static mp_limb_t
                    142: #if __STDC__
                    143: mpn_bz_divrem_aux (mp_ptr qp, mp_ptr np, mp_srcptr dp, mp_size_t n, mp_ptr tmp)
                    144: #else
                    145: mpn_bz_divrem_aux (qp, np, dp, n, tmp)
                    146:      mp_ptr qp;
                    147:      mp_ptr np;
                    148:      mp_srcptr dp;
                    149:      mp_size_t n;
                    150:      mp_ptr tmp;
                    151: #endif
                    152: {
                    153:   mp_limb_t qhl;
                    154:
                    155:   if (n % 2 != 0)
                    156:     {
                    157:       /* divide (2n - 2) most significant limbs from np by those (n - 1) from dp */
                    158:       qhl = mpn_bz_divrem_aux (qp + 1, np + 2, dp + 1, n - 1, tmp);
                    159:       /* now (qp + 1, n - 1) contains the quotient of (np + 2, 2n - 2) by
                    160:         (dp + 1, n - 1) and (np + 2, n - 1) contains the remainder */
                    161:       if (mpn_sub_1 (np + n, np + n, 1,
                    162:                     mpn_submul_1 (np + 1, qp + 1, n - 1, dp[0])))
                    163:        {
                    164:          /* quotient too large */
                    165:          qhl -= mpn_sub_1 (qp + 1, qp + 1, n - 1, 1L);
                    166:          if (mpn_add_n (np + 1, np + 1, dp, n) == 0)
                    167:            { /* still too large */
                    168:              qhl -= mpn_sub_1 (qp + 1, qp + 1, n - 1, 1L);
                    169:              mpn_add_n (np + 1, np + 1, dp, n); /* always carry here */
                    170:            }
                    171:        }
                    172:       /* now divide (np, n + 1) by (dp, n) */
                    173:       qhl += mpn_add_1 (qp + 1, qp + 1, n - 1,
                    174:                        mpn_sb_divrem_mn (qp, np, n + 1, dp, n));
                    175:     }
                    176:   else
                    177:     {
                    178:       mp_size_t n2 = n/2;
                    179:       qhl = mpn_bz_div_3_halves_by_2 (qp + n2, np + n2, dp, n2, tmp);
                    180:       qhl += mpn_add_1 (qp + n2, qp + n2, n2,
                    181:                       mpn_bz_div_3_halves_by_2 (qp, np, dp, n2, tmp));
                    182:     }
                    183:   return qhl;
                    184: }
                    185:
                    186: /* divides (np, 3n) by (dp, 2n) and puts the quotient in (qp, n),
                    187:    the remainder in (np, 2n) */
                    188:
                    189: static mp_limb_t
                    190: #if __STDC__
                    191: mpn_bz_div_3_halves_by_2 (mp_ptr qp, mp_ptr np, mp_srcptr dp, mp_size_t n,
                    192:                           mp_ptr tmp)
                    193: #else
                    194: mpn_bz_div_3_halves_by_2 (qp, np, dp, n, tmp)
                    195:      mp_ptr qp;
                    196:      mp_ptr np;
                    197:      mp_srcptr dp;
                    198:      mp_size_t n;
                    199:      mp_ptr tmp;
                    200: #endif
                    201: {
                    202:   mp_size_t twon = n + n;
                    203:   mp_limb_t qhl;
                    204:
                    205:   if (n < BZ_THRESHOLD)
                    206:     qhl = mpn_sb_divrem_mn (qp, np + n, twon, dp + n, n);
                    207:   else
                    208:     qhl = mpn_bz_divrem_aux (qp, np + n, dp + n, n, tmp);
                    209:   /* q = (qp, n), c = (np + n, n) with the notations of [1] */
                    210:   mpn_mul_n (tmp, qp, dp, n);
                    211:   if (qhl != 0)
                    212:     mpn_add_n (tmp + n, tmp + n, dp, n);
                    213:   if (mpn_sub_n (np, np, tmp, twon))           /* R = (np, 2n) */
                    214:     {
                    215:       qhl -= mpn_sub_1 (qp, qp, n, 1L);
                    216:       if (mpn_add_n (np, np, dp, twon) == 0)
                    217:        { /* qp still too large */
                    218:          qhl -= mpn_sub_1 (qp, qp, n, 1L);
                    219:          mpn_add_n (np, np, dp, twon);         /* always carry here */
                    220:        }
                    221:     }
                    222:   return qhl;
                    223: }

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>