xref: /dragonfly/contrib/gmp/mpn/generic/toom52_mul.c (revision 25a2db75)
1 /* mpn_toom52_mul -- Multiply {ap,an} and {bp,bn} where an is nominally 4/3
2    times as large as bn.  Or more accurately, bn < an < 2 bn.
3 
4    Contributed to the GNU project by Marco Bodrato.
5 
6    The idea of applying toom to unbalanced multiplication is due to Marco
7    Bodrato and Alberto Zanoni.
8 
9    THE FUNCTION IN THIS FILE IS INTERNAL WITH A MUTABLE INTERFACE.  IT IS ONLY
10    SAFE TO REACH IT THROUGH DOCUMENTED INTERFACES.  IN FACT, IT IS ALMOST
11    GUARANTEED THAT IT WILL CHANGE OR DISAPPEAR IN A FUTURE GNU MP RELEASE.
12 
13 Copyright 2009 Free Software Foundation, Inc.
14 
15 This file is part of the GNU MP Library.
16 
17 The GNU MP Library is free software; you can redistribute it and/or modify
18 it under the terms of the GNU Lesser General Public License as published by
19 the Free Software Foundation; either version 3 of the License, or (at your
20 option) any later version.
21 
22 The GNU MP Library is distributed in the hope that it will be useful, but
23 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
24 or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
25 License for more details.
26 
27 You should have received a copy of the GNU Lesser General Public License
28 along with the GNU MP Library.  If not, see http://www.gnu.org/licenses/.  */
29 
30 
31 #include "gmp.h"
32 #include "gmp-impl.h"
33 
34 /* Evaluate in: -2, -1, 0, +1, +2, +inf
35 
36   <-s-><--n--><--n--><--n--><--n-->
37    ___ ______ ______ ______ ______
38   |a4_|___a3_|___a2_|___a1_|___a0_|
39 			|b1|___b0_|
40 			<t-><--n-->
41 
42   v0  =  a0                  * b0      #   A(0)*B(0)
43   v1  = (a0+ a1+ a2+ a3+  a4)*(b0+ b1) #   A(1)*B(1)      ah  <= 4   bh <= 1
44   vm1 = (a0- a1+ a2- a3+  a4)*(b0- b1) #  A(-1)*B(-1)    |ah| <= 2   bh  = 0
45   v2  = (a0+2a1+4a2+8a3+16a4)*(b0+2b1) #   A(2)*B(2)      ah  <= 30  bh <= 2
46   vm2 = (a0-2a1+4a2-8a3+16a4)*(b0-2b1) #  A(-2)*B(-2)    |ah| <= 20 |bh|<= 1
47   vinf=                   a4 *     b1  # A(inf)*B(inf)
48 
49   Some slight optimization in evaluation are taken from the paper:
50   "Towards Optimal Toom-Cook Multiplication for Univariate and
51   Multivariate Polynomials in Characteristic 2 and 0."
52 */
53 
54 void
55 mpn_toom52_mul (mp_ptr pp,
56 		mp_srcptr ap, mp_size_t an,
57 		mp_srcptr bp, mp_size_t bn, mp_ptr scratch)
58 {
59   mp_size_t n, s, t;
60   enum toom6_flags flags;
61 
62 #define a0  ap
63 #define a1  (ap + n)
64 #define a2  (ap + 2 * n)
65 #define a3  (ap + 3 * n)
66 #define a4  (ap + 4 * n)
67 #define b0  bp
68 #define b1  (bp + n)
69 
70   n = 1 + (2 * an >= 5 * bn ? (an - 1) / (size_t) 5 : (bn - 1) >> 1);
71 
72   s = an - 4 * n;
73   t = bn - n;
74 
75   ASSERT (0 < s && s <= n);
76   ASSERT (0 < t && t <= n);
77 
78   /* Ensures that 5 values of n+1 limbs each fits in the product area.
79      Borderline cases are an = 32, bn = 8, n = 7, and an = 36, bn = 9,
80      n = 8. */
81   ASSERT (s+t >= 5);
82 
83 #define v0    pp				/* 2n */
84 #define vm1   (scratch)				/* 2n+1 */
85 #define v1    (pp + 2 * n)			/* 2n+1 */
86 #define vm2   (scratch + 2 * n + 1)		/* 2n+1 */
87 #define v2    (scratch + 4 * n + 2)		/* 2n+1 */
88 #define vinf  (pp + 5 * n)			/* s+t */
89 #define bs1    pp				/* n+1 */
90 #define bsm1  (scratch + 2 * n + 2)		/* n   */
91 #define asm1  (scratch + 3 * n + 3)		/* n+1 */
92 #define asm2  (scratch + 4 * n + 4)		/* n+1 */
93 #define bsm2  (pp + n + 1)			/* n+1 */
94 #define bs2   (pp + 2 * n + 2)			/* n+1 */
95 #define as2   (pp + 3 * n + 3)			/* n+1 */
96 #define as1   (pp + 4 * n + 4)			/* n+1 */
97 
98   /* Scratch need is 6 * n + 3 + 1. We need one extra limb, because
99      products will overwrite 2n+2 limbs. */
100 
101 #define a0a2  scratch
102 #define a1a3  asm1
103 
104   /* Compute as2 and asm2.  */
105   flags = toom6_vm2_neg & mpn_toom_eval_pm2 (as2, asm2, 4, ap, n, s, a1a3);
106 
107   /* Compute bs1 and bsm1.  */
108   if (t == n)
109     {
110 #if HAVE_NATIVE_mpn_add_n_sub_n
111       mp_limb_t cy;
112 
113       if (mpn_cmp (b0, b1, n) < 0)
114 	{
115 	  cy = mpn_add_n_sub_n (bs1, bsm1, b1, b0, n);
116 	  flags ^= toom6_vm1_neg;
117 	}
118       else
119 	{
120 	  cy = mpn_add_n_sub_n (bs1, bsm1, b0, b1, n);
121 	}
122       bs1[n] = cy >> 1;
123 #else
124       bs1[n] = mpn_add_n (bs1, b0, b1, n);
125       if (mpn_cmp (b0, b1, n) < 0)
126 	{
127 	  mpn_sub_n (bsm1, b1, b0, n);
128 	  flags ^= toom6_vm1_neg;
129 	}
130       else
131 	{
132 	  mpn_sub_n (bsm1, b0, b1, n);
133 	}
134 #endif
135     }
136   else
137     {
138       bs1[n] = mpn_add (bs1, b0, n, b1, t);
139       if (mpn_zero_p (b0 + t, n - t) && mpn_cmp (b0, b1, t) < 0)
140 	{
141 	  mpn_sub_n (bsm1, b1, b0, t);
142 	  MPN_ZERO (bsm1 + t, n - t);
143 	  flags ^= toom6_vm1_neg;
144 	}
145       else
146 	{
147 	  mpn_sub (bsm1, b0, n, b1, t);
148 	}
149     }
150 
151   /* Compute bs2 and bsm2, recycling bs1 and bsm1. bs2=bs1+b1; bsm2=bsm1-b1  */
152   mpn_add (bs2, bs1, n+1, b1, t);
153   if (flags & toom6_vm1_neg )
154     {
155       bsm2[n] = mpn_add (bsm2, bsm1, n, b1, t);
156       flags ^= toom6_vm2_neg;
157     }
158   else
159     {
160       bsm2[n] = 0;
161       if (t == n)
162 	{
163 	  if (mpn_cmp (bsm1, b1, n) < 0)
164 	    {
165 	      mpn_sub_n (bsm2, b1, bsm1, n);
166 	      flags ^= toom6_vm2_neg;
167 	    }
168 	  else
169 	    {
170 	      mpn_sub_n (bsm2, bsm1, b1, n);
171 	    }
172 	}
173       else
174 	{
175 	  if (mpn_zero_p (bsm1 + t, n - t) && mpn_cmp (bsm1, b1, t) < 0)
176 	    {
177 	      mpn_sub_n (bsm2, b1, bsm1, t);
178 	      MPN_ZERO (bsm2 + t, n - t);
179 	      flags ^= toom6_vm2_neg;
180 	    }
181 	  else
182 	    {
183 	      mpn_sub (bsm2, bsm1, n, b1, t);
184 	    }
185 	}
186     }
187 
188   /* Compute as1 and asm1.  */
189   flags ^= toom6_vm1_neg & mpn_toom_eval_pm1 (as1, asm1, 4, ap, n, s, a0a2);
190 
191   ASSERT (as1[n] <= 4);
192   ASSERT (bs1[n] <= 1);
193   ASSERT (asm1[n] <= 2);
194 /*   ASSERT (bsm1[n] <= 1); */
195   ASSERT (as2[n] <=30);
196   ASSERT (bs2[n] <= 2);
197   ASSERT (asm2[n] <= 20);
198   ASSERT (bsm2[n] <= 1);
199 
200   /* vm1, 2n+1 limbs */
201   mpn_mul (vm1, asm1, n+1, bsm1, n);  /* W4 */
202 
203   /* vm2, 2n+1 limbs */
204   mpn_mul_n (vm2, asm2, bsm2, n+1);  /* W2 */
205 
206   /* v2, 2n+1 limbs */
207   mpn_mul_n (v2, as2, bs2, n+1);  /* W1 */
208 
209   /* v1, 2n+1 limbs */
210   mpn_mul_n (v1, as1, bs1, n+1);  /* W3 */
211 
212   /* vinf, s+t limbs */   /* W0 */
213   if (s > t)  mpn_mul (vinf, a4, s, b1, t);
214   else        mpn_mul (vinf, b1, t, a4, s);
215 
216   /* v0, 2n limbs */
217   mpn_mul_n (v0, ap, bp, n);  /* W5 */
218 
219   mpn_toom_interpolate_6pts (pp, n, flags, vm1, vm2, v2, t + s);
220 
221 #undef v0
222 #undef vm1
223 #undef v1
224 #undef vm2
225 #undef v2
226 #undef vinf
227 #undef bs1
228 #undef bs2
229 #undef bsm1
230 #undef bsm2
231 #undef asm1
232 #undef asm2
233 #undef as1
234 #undef as2
235 #undef a0a2
236 #undef b0b2
237 #undef a1a3
238 #undef a0
239 #undef a1
240 #undef a2
241 #undef a3
242 #undef b0
243 #undef b1
244 #undef b2
245 
246 }
247