Remove some redundant computations in s_sin.c
[glibc.git] / stdlib / mul_n.c
blobf0a9a304dd07c435fc1fd2d5649a457aa74c7ae9
1 /* mpn_mul_n -- Multiply two natural numbers of length n.
3 Copyright (C) 1991-2013 Free Software Foundation, Inc.
5 This file is part of the GNU MP Library.
7 The GNU MP Library is free software; you can redistribute it and/or modify
8 it under the terms of the GNU Lesser General Public License as published by
9 the Free Software Foundation; either version 2.1 of the License, or (at your
10 option) any later version.
12 The GNU MP Library is distributed in the hope that it will be useful, but
13 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
14 or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
15 License for more details.
17 You should have received a copy of the GNU Lesser General Public License
18 along with the GNU MP Library; see the file COPYING.LIB. If not, see
19 <http://www.gnu.org/licenses/>. */
21 #include <gmp.h>
22 #include "gmp-impl.h"
24 /* Multiply the natural numbers u (pointed to by UP) and v (pointed to by VP),
25 both with SIZE limbs, and store the result at PRODP. 2 * SIZE limbs are
26 always stored. Return the most significant limb.
28 Argument constraints:
29 1. PRODP != UP and PRODP != VP, i.e. the destination
30 must be distinct from the multiplier and the multiplicand. */
32 /* If KARATSUBA_THRESHOLD is not already defined, define it to a
33 value which is good on most machines. */
34 #ifndef KARATSUBA_THRESHOLD
35 #define KARATSUBA_THRESHOLD 32
36 #endif
38 /* The code can't handle KARATSUBA_THRESHOLD smaller than 2. */
39 #if KARATSUBA_THRESHOLD < 2
40 #undef KARATSUBA_THRESHOLD
41 #define KARATSUBA_THRESHOLD 2
42 #endif
44 /* Handle simple cases with traditional multiplication.
46 This is the most critical code of multiplication. All multiplies rely
47 on this, both small and huge. Small ones arrive here immediately. Huge
48 ones arrive here as this is the base case for Karatsuba's recursive
49 algorithm below. */
51 void
52 #if __STDC__
53 impn_mul_n_basecase (mp_ptr prodp, mp_srcptr up, mp_srcptr vp, mp_size_t size)
54 #else
55 impn_mul_n_basecase (prodp, up, vp, size)
56 mp_ptr prodp;
57 mp_srcptr up;
58 mp_srcptr vp;
59 mp_size_t size;
60 #endif
62 mp_size_t i;
63 mp_limb_t cy_limb;
64 mp_limb_t v_limb;
66 /* Multiply by the first limb in V separately, as the result can be
67 stored (not added) to PROD. We also avoid a loop for zeroing. */
68 v_limb = vp[0];
69 if (v_limb <= 1)
71 if (v_limb == 1)
72 MPN_COPY (prodp, up, size);
73 else
74 MPN_ZERO (prodp, size);
75 cy_limb = 0;
77 else
78 cy_limb = mpn_mul_1 (prodp, up, size, v_limb);
80 prodp[size] = cy_limb;
81 prodp++;
83 /* For each iteration in the outer loop, multiply one limb from
84 U with one limb from V, and add it to PROD. */
85 for (i = 1; i < size; i++)
87 v_limb = vp[i];
88 if (v_limb <= 1)
90 cy_limb = 0;
91 if (v_limb == 1)
92 cy_limb = mpn_add_n (prodp, prodp, up, size);
94 else
95 cy_limb = mpn_addmul_1 (prodp, up, size, v_limb);
97 prodp[size] = cy_limb;
98 prodp++;
102 void
103 #if __STDC__
104 impn_mul_n (mp_ptr prodp,
105 mp_srcptr up, mp_srcptr vp, mp_size_t size, mp_ptr tspace)
106 #else
107 impn_mul_n (prodp, up, vp, size, tspace)
108 mp_ptr prodp;
109 mp_srcptr up;
110 mp_srcptr vp;
111 mp_size_t size;
112 mp_ptr tspace;
113 #endif
115 if ((size & 1) != 0)
117 /* The size is odd, the code code below doesn't handle that.
118 Multiply the least significant (size - 1) limbs with a recursive
119 call, and handle the most significant limb of S1 and S2
120 separately. */
121 /* A slightly faster way to do this would be to make the Karatsuba
122 code below behave as if the size were even, and let it check for
123 odd size in the end. I.e., in essence move this code to the end.
124 Doing so would save us a recursive call, and potentially make the
125 stack grow a lot less. */
127 mp_size_t esize = size - 1; /* even size */
128 mp_limb_t cy_limb;
130 MPN_MUL_N_RECURSE (prodp, up, vp, esize, tspace);
131 cy_limb = mpn_addmul_1 (prodp + esize, up, esize, vp[esize]);
132 prodp[esize + esize] = cy_limb;
133 cy_limb = mpn_addmul_1 (prodp + esize, vp, size, up[esize]);
135 prodp[esize + size] = cy_limb;
137 else
139 /* Anatolij Alekseevich Karatsuba's divide-and-conquer algorithm.
141 Split U in two pieces, U1 and U0, such that
142 U = U0 + U1*(B**n),
143 and V in V1 and V0, such that
144 V = V0 + V1*(B**n).
146 UV is then computed recursively using the identity
148 2n n n n
149 UV = (B + B )U V + B (U -U )(V -V ) + (B + 1)U V
150 1 1 1 0 0 1 0 0
152 Where B = 2**BITS_PER_MP_LIMB. */
154 mp_size_t hsize = size >> 1;
155 mp_limb_t cy;
156 int negflg;
158 /*** Product H. ________________ ________________
159 |_____U1 x V1____||____U0 x V0_____| */
160 /* Put result in upper part of PROD and pass low part of TSPACE
161 as new TSPACE. */
162 MPN_MUL_N_RECURSE (prodp + size, up + hsize, vp + hsize, hsize, tspace);
164 /*** Product M. ________________
165 |_(U1-U0)(V0-V1)_| */
166 if (mpn_cmp (up + hsize, up, hsize) >= 0)
168 mpn_sub_n (prodp, up + hsize, up, hsize);
169 negflg = 0;
171 else
173 mpn_sub_n (prodp, up, up + hsize, hsize);
174 negflg = 1;
176 if (mpn_cmp (vp + hsize, vp, hsize) >= 0)
178 mpn_sub_n (prodp + hsize, vp + hsize, vp, hsize);
179 negflg ^= 1;
181 else
183 mpn_sub_n (prodp + hsize, vp, vp + hsize, hsize);
184 /* No change of NEGFLG. */
186 /* Read temporary operands from low part of PROD.
187 Put result in low part of TSPACE using upper part of TSPACE
188 as new TSPACE. */
189 MPN_MUL_N_RECURSE (tspace, prodp, prodp + hsize, hsize, tspace + size);
191 /*** Add/copy product H. */
192 MPN_COPY (prodp + hsize, prodp + size, hsize);
193 cy = mpn_add_n (prodp + size, prodp + size, prodp + size + hsize, hsize);
195 /*** Add product M (if NEGFLG M is a negative number). */
196 if (negflg)
197 cy -= mpn_sub_n (prodp + hsize, prodp + hsize, tspace, size);
198 else
199 cy += mpn_add_n (prodp + hsize, prodp + hsize, tspace, size);
201 /*** Product L. ________________ ________________
202 |________________||____U0 x V0_____| */
203 /* Read temporary operands from low part of PROD.
204 Put result in low part of TSPACE using upper part of TSPACE
205 as new TSPACE. */
206 MPN_MUL_N_RECURSE (tspace, up, vp, hsize, tspace + size);
208 /*** Add/copy Product L (twice). */
210 cy += mpn_add_n (prodp + hsize, prodp + hsize, tspace, size);
211 if (cy)
212 mpn_add_1 (prodp + hsize + size, prodp + hsize + size, hsize, cy);
214 MPN_COPY (prodp, tspace, hsize);
215 cy = mpn_add_n (prodp + hsize, prodp + hsize, tspace + hsize, hsize);
216 if (cy)
217 mpn_add_1 (prodp + size, prodp + size, size, 1);
221 void
222 #if __STDC__
223 impn_sqr_n_basecase (mp_ptr prodp, mp_srcptr up, mp_size_t size)
224 #else
225 impn_sqr_n_basecase (prodp, up, size)
226 mp_ptr prodp;
227 mp_srcptr up;
228 mp_size_t size;
229 #endif
231 mp_size_t i;
232 mp_limb_t cy_limb;
233 mp_limb_t v_limb;
235 /* Multiply by the first limb in V separately, as the result can be
236 stored (not added) to PROD. We also avoid a loop for zeroing. */
237 v_limb = up[0];
238 if (v_limb <= 1)
240 if (v_limb == 1)
241 MPN_COPY (prodp, up, size);
242 else
243 MPN_ZERO (prodp, size);
244 cy_limb = 0;
246 else
247 cy_limb = mpn_mul_1 (prodp, up, size, v_limb);
249 prodp[size] = cy_limb;
250 prodp++;
252 /* For each iteration in the outer loop, multiply one limb from
253 U with one limb from V, and add it to PROD. */
254 for (i = 1; i < size; i++)
256 v_limb = up[i];
257 if (v_limb <= 1)
259 cy_limb = 0;
260 if (v_limb == 1)
261 cy_limb = mpn_add_n (prodp, prodp, up, size);
263 else
264 cy_limb = mpn_addmul_1 (prodp, up, size, v_limb);
266 prodp[size] = cy_limb;
267 prodp++;
271 void
272 #if __STDC__
273 impn_sqr_n (mp_ptr prodp,
274 mp_srcptr up, mp_size_t size, mp_ptr tspace)
275 #else
276 impn_sqr_n (prodp, up, size, tspace)
277 mp_ptr prodp;
278 mp_srcptr up;
279 mp_size_t size;
280 mp_ptr tspace;
281 #endif
283 if ((size & 1) != 0)
285 /* The size is odd, the code code below doesn't handle that.
286 Multiply the least significant (size - 1) limbs with a recursive
287 call, and handle the most significant limb of S1 and S2
288 separately. */
289 /* A slightly faster way to do this would be to make the Karatsuba
290 code below behave as if the size were even, and let it check for
291 odd size in the end. I.e., in essence move this code to the end.
292 Doing so would save us a recursive call, and potentially make the
293 stack grow a lot less. */
295 mp_size_t esize = size - 1; /* even size */
296 mp_limb_t cy_limb;
298 MPN_SQR_N_RECURSE (prodp, up, esize, tspace);
299 cy_limb = mpn_addmul_1 (prodp + esize, up, esize, up[esize]);
300 prodp[esize + esize] = cy_limb;
301 cy_limb = mpn_addmul_1 (prodp + esize, up, size, up[esize]);
303 prodp[esize + size] = cy_limb;
305 else
307 mp_size_t hsize = size >> 1;
308 mp_limb_t cy;
310 /*** Product H. ________________ ________________
311 |_____U1 x U1____||____U0 x U0_____| */
312 /* Put result in upper part of PROD and pass low part of TSPACE
313 as new TSPACE. */
314 MPN_SQR_N_RECURSE (prodp + size, up + hsize, hsize, tspace);
316 /*** Product M. ________________
317 |_(U1-U0)(U0-U1)_| */
318 if (mpn_cmp (up + hsize, up, hsize) >= 0)
320 mpn_sub_n (prodp, up + hsize, up, hsize);
322 else
324 mpn_sub_n (prodp, up, up + hsize, hsize);
327 /* Read temporary operands from low part of PROD.
328 Put result in low part of TSPACE using upper part of TSPACE
329 as new TSPACE. */
330 MPN_SQR_N_RECURSE (tspace, prodp, hsize, tspace + size);
332 /*** Add/copy product H. */
333 MPN_COPY (prodp + hsize, prodp + size, hsize);
334 cy = mpn_add_n (prodp + size, prodp + size, prodp + size + hsize, hsize);
336 /*** Add product M (if NEGFLG M is a negative number). */
337 cy -= mpn_sub_n (prodp + hsize, prodp + hsize, tspace, size);
339 /*** Product L. ________________ ________________
340 |________________||____U0 x U0_____| */
341 /* Read temporary operands from low part of PROD.
342 Put result in low part of TSPACE using upper part of TSPACE
343 as new TSPACE. */
344 MPN_SQR_N_RECURSE (tspace, up, hsize, tspace + size);
346 /*** Add/copy Product L (twice). */
348 cy += mpn_add_n (prodp + hsize, prodp + hsize, tspace, size);
349 if (cy)
350 mpn_add_1 (prodp + hsize + size, prodp + hsize + size, hsize, cy);
352 MPN_COPY (prodp, tspace, hsize);
353 cy = mpn_add_n (prodp + hsize, prodp + hsize, tspace + hsize, hsize);
354 if (cy)
355 mpn_add_1 (prodp + size, prodp + size, size, 1);
359 /* This should be made into an inline function in gmp.h. */
360 void
361 #if __STDC__
362 mpn_mul_n (mp_ptr prodp, mp_srcptr up, mp_srcptr vp, mp_size_t size)
363 #else
364 mpn_mul_n (prodp, up, vp, size)
365 mp_ptr prodp;
366 mp_srcptr up;
367 mp_srcptr vp;
368 mp_size_t size;
369 #endif
371 TMP_DECL (marker);
372 TMP_MARK (marker);
373 if (up == vp)
375 if (size < KARATSUBA_THRESHOLD)
377 impn_sqr_n_basecase (prodp, up, size);
379 else
381 mp_ptr tspace;
382 tspace = (mp_ptr) TMP_ALLOC (2 * size * BYTES_PER_MP_LIMB);
383 impn_sqr_n (prodp, up, size, tspace);
386 else
388 if (size < KARATSUBA_THRESHOLD)
390 impn_mul_n_basecase (prodp, up, vp, size);
392 else
394 mp_ptr tspace;
395 tspace = (mp_ptr) TMP_ALLOC (2 * size * BYTES_PER_MP_LIMB);
396 impn_mul_n (prodp, up, vp, size, tspace);
399 TMP_FREE (marker);