Coverage Report

Created: 2026-03-31 06:37

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/gmp/mpn/toom22_mul.c
Line
Count
Source
1
/* mpn_toom22_mul -- Multiply {ap,an} and {bp,bn} where an >= bn.  Or more
2
   accurately, bn <= an < 2bn.
3
4
   Contributed to the GNU project by Torbjorn Granlund.
5
6
   THE FUNCTION IN THIS FILE IS INTERNAL WITH A MUTABLE INTERFACE.  IT IS ONLY
7
   SAFE TO REACH IT THROUGH DOCUMENTED INTERFACES.  IN FACT, IT IS ALMOST
8
   GUARANTEED THAT IT WILL CHANGE OR DISAPPEAR IN A FUTURE GNU MP RELEASE.
9
10
Copyright 2006-2010, 2012, 2014, 2018, 2020 Free Software Foundation, Inc.
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 either:
16
17
  * the GNU Lesser General Public License as published by the Free
18
    Software Foundation; either version 3 of the License, or (at your
19
    option) any later version.
20
21
or
22
23
  * the GNU General Public License as published by the Free Software
24
    Foundation; either version 2 of the License, or (at your option) any
25
    later version.
26
27
or both in parallel, as here.
28
29
The GNU MP Library is distributed in the hope that it will be useful, but
30
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
31
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
32
for more details.
33
34
You should have received copies of the GNU General Public License and the
35
GNU Lesser General Public License along with the GNU MP Library.  If not,
36
see https://www.gnu.org/licenses/.  */
37
38
39
#include "gmp-impl.h"
40
41
/* Evaluate in: -1, 0, +inf
42
43
  <-s--><--n-->
44
   ____ ______
45
  |_a1_|___a0_|
46
   |b1_|___b0_|
47
   <-t-><--n-->
48
49
  v0  =  a0     * b0       #   A(0)*B(0)
50
  vm1 = (a0- a1)*(b0- b1)  #  A(-1)*B(-1)
51
  vinf=      a1 *     b1   # A(inf)*B(inf)
52
*/
53
54
#if TUNE_PROGRAM_BUILD || WANT_FAT_BINARY
55
#define MAYBE_mul_toom22   1
56
#else
57
#define MAYBE_mul_toom22            \
58
214M
  (MUL_TOOM33_THRESHOLD >= 2 * MUL_TOOM22_THRESHOLD)
59
#endif
60
61
#define TOOM22_MUL_N_REC(p, a, b, n, ws)        \
62
107M
  do {                 \
63
107M
    if (! MAYBE_mul_toom22            \
64
107M
  || BELOW_THRESHOLD (n, MUL_TOOM22_THRESHOLD))     \
65
107M
      mpn_mul_basecase (p, a, n, b, n);         \
66
107M
    else                \
67
107M
      mpn_toom22_mul (p, a, n, b, n, ws);       \
68
107M
  } while (0)
69
70
/* Normally, this calls mul_basecase or toom22_mul.  But when when the fraction
71
   MUL_TOOM33_THRESHOLD / MUL_TOOM22_THRESHOLD is large, an initially small
72
   relative unbalance will become a larger and larger relative unbalance with
73
   each recursion (the difference s-t will be invariant over recursive calls).
74
   Therefore, we need to call toom32_mul.  FIXME: Suppress depending on
75
   MUL_TOOM33_THRESHOLD / MUL_TOOM22_THRESHOLD and on MUL_TOOM22_THRESHOLD.  */
76
#define TOOM22_MUL_REC(p, a, an, b, bn, ws)       \
77
6.37k
  do {                 \
78
6.37k
    if (! MAYBE_mul_toom22            \
79
6.37k
  || BELOW_THRESHOLD (bn, MUL_TOOM22_THRESHOLD))      \
80
6.37k
      mpn_mul_basecase (p, a, an, b, bn);       \
81
6.37k
    else if (4 * an < 5 * bn)           \
82
504
      mpn_toom22_mul (p, a, an, b, bn, ws);       \
83
504
    else                \
84
504
      mpn_toom32_mul (p, a, an, b, bn, ws);       \
85
6.37k
  } while (0)
86
87
void
88
mpn_toom22_mul (mp_ptr pp,
89
    mp_srcptr ap, mp_size_t an,
90
    mp_srcptr bp, mp_size_t bn,
91
    mp_ptr scratch)
92
35.6M
{
93
35.6M
  const int __gmpn_cpuvec_initialized = 1;
94
35.6M
  mp_size_t n, s, t;
95
35.6M
  int vm1_neg;
96
35.6M
  mp_limb_t cy, cy2;
97
35.6M
  mp_ptr asm1;
98
35.6M
  mp_ptr bsm1;
99
100
85.2M
#define a0  ap
101
58.3M
#define a1  (ap + n)
102
73.2M
#define b0  bp
103
59.7M
#define b1  (bp + n)
104
105
35.6M
  s = an >> 1;
106
35.6M
  n = an - s;
107
35.6M
  t = bn - n;
108
109
35.6M
  ASSERT (an >= bn);
110
111
35.6M
  ASSERT (0 < s && s <= n && (n - s) == (an & 1));
112
35.6M
  ASSERT (0 < t && t <= s);
113
114
35.6M
  asm1 = pp;
115
35.6M
  bsm1 = pp + n;
116
117
35.6M
  vm1_neg = 0;
118
119
  /* Compute asm1.  */
120
35.6M
  if ((an & 1) == 0) /* s == n */
121
22.1M
    {
122
22.1M
      if (mpn_cmp (a0, a1, n) < 0)
123
9.24M
  {
124
9.24M
    mpn_sub_n (asm1, a1, a0, n);
125
9.24M
    vm1_neg = 1;
126
9.24M
  }
127
12.9M
      else
128
12.9M
  {
129
12.9M
    mpn_sub_n (asm1, a0, a1, n);
130
12.9M
  }
131
22.1M
    }
132
13.5M
  else /* n - s == 1 */
133
13.5M
    {
134
13.5M
      if (a0[s] == 0 && mpn_cmp (a0, a1, s) < 0)
135
195k
  {
136
195k
    mpn_sub_n (asm1, a1, a0, s);
137
195k
    asm1[s] = 0;
138
195k
    vm1_neg = 1;
139
195k
  }
140
13.3M
      else
141
13.3M
  {
142
13.3M
    asm1[s] = a0[s] - mpn_sub_n (asm1, a0, a1, s);
143
13.3M
  }
144
13.5M
    }
145
146
  /* Compute bsm1.  */
147
35.6M
  if (t == n)
148
22.1M
    {
149
22.1M
      if (mpn_cmp (b0, b1, n) < 0)
150
10.4M
  {
151
10.4M
    mpn_sub_n (bsm1, b1, b0, n);
152
10.4M
    vm1_neg ^= 1;
153
10.4M
  }
154
11.7M
      else
155
11.7M
  {
156
11.7M
    mpn_sub_n (bsm1, b0, b1, n);
157
11.7M
  }
158
22.1M
    }
159
13.5M
  else
160
13.5M
    {
161
13.5M
      if (mpn_zero_p (b0 + t, n - t) && mpn_cmp (b0, b1, t) < 0)
162
969k
  {
163
969k
    mpn_sub_n (bsm1, b1, b0, t);
164
969k
    MPN_ZERO (bsm1 + t, n - t);
165
969k
    vm1_neg ^= 1;
166
969k
  }
167
12.5M
      else
168
12.5M
  {
169
12.5M
    mpn_sub (bsm1, b0, n, b1, t);
170
12.5M
  }
171
13.5M
    }
172
173
71.3M
#define v0  pp        /* 2n */
174
71.3M
#define vinf  (pp + 2 * n)      /* s+t */
175
35.6M
#define vm1 scratch        /* 2n */
176
35.6M
#define scratch_out scratch + 2 * n
177
178
  /* vm1, 2n limbs */
179
35.6M
  TOOM22_MUL_N_REC (vm1, asm1, bsm1, n, scratch_out);
180
181
35.6M
  if (s > t)  TOOM22_MUL_REC (vinf, a1, s, b1, t, scratch_out);
182
35.6M
  else        TOOM22_MUL_N_REC (vinf, a1, b1, s, scratch_out);
183
184
  /* v0, 2n limbs */
185
35.6M
  TOOM22_MUL_N_REC (v0, ap, bp, n, scratch_out);
186
187
  /* H(v0) + L(vinf) */
188
35.6M
  cy = mpn_add_n (pp + 2 * n, v0 + n, vinf, n);
189
190
  /* L(v0) + (H(v0) + L(vinf)) */
191
35.6M
  cy2 = cy + mpn_add_n (pp + n, pp + 2 * n, v0, n);
192
193
  /* (H(v0) + L(vinf)) + H(vinf) */
194
35.6M
  cy += mpn_add (pp + 2 * n, pp + 2 * n, n, vinf + n, s + t - n);
195
196
35.6M
  if (vm1_neg)
197
10.6M
    cy += mpn_add_n (pp + n, pp + n, vm1, 2 * n);
198
25.0M
  else {
199
25.0M
    cy -= mpn_sub_n (pp + n, pp + n, vm1, 2 * n);
200
25.0M
    if (UNLIKELY (cy + 1 == 0)) { /* cy is negative */
201
      /* The total contribution of v0+vinf-vm1 can not be negative. */
202
#if WANT_ASSERT
203
      /* The borrow in cy stops the propagation of the carry cy2, */
204
      ASSERT (cy2 == 1);
205
      cy += mpn_add_1 (pp + 2 * n, pp + 2 * n, n, cy2);
206
      ASSERT (cy == 0);
207
#else
208
      /* we simply fill the area with zeros. */
209
3.01k
      MPN_FILL (pp + 2 * n, n, 0);
210
      /* ASSERT (s + t == n || mpn_zero_p (pp + 3 * n, s + t - n)); */
211
3.01k
#endif
212
3.01k
      return;
213
3.01k
    }
214
25.0M
  }
215
216
35.6M
  ASSERT (cy  <= 2);
217
35.6M
  ASSERT (cy2 <= 2);
218
219
35.6M
  MPN_INCR_U (pp + 2 * n, s + t, cy2);
220
  /* if s+t==n, cy is zero, but we should not access pp[3*n] at all. */
221
35.6M
  MPN_INCR_U (pp + 3 * n, s + t - n, cy);
222
35.6M
}