LMMP 4.3.0
Lamina High-Precision Arithmetic Library
载入中...
搜索中...
未找到
mul_toom_interp5.c
浏览该文件的文档.
1/**
2 * Copyright (C) 2026 HJimmyK (Jericho Knox)
3 *
4 * This file is part of LMMP.
5 *
6 * LMMP is free software: you can redistribute it and/or modify it
7 * under the terms of the GNU Lesser General Public License as
8 * published by the Free Software Foundation; either version 3 of
9 * the License, or (at your option) any later version.
10 *
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library. If not, see
18 * <https://www.gnu.org/licenses/>.
19 *
20 * SPDX-License-Identifier: LGPL-3.0-or-later
21 */
22
23#include "lmmpn/divexact.h"
24#include "lmmpn/toom_interp.h"
25
26
27void lmmp_toom_interp5_(mp_ptr dst, mp_ptr v2, mp_ptr vm1, mp_size_t n, mp_size_t spt, int vm1_neg, mp_limb_t vinf0) {
28 mp_limb_t cy, saved;
29 mp_size_t dnp = 2 * n + 1;
30
31#define r0 (dst)
32#define r1 (dst + n)
33#define r2 (dst + 2 * n)
34#define r3 (dst + 3 * n)
35#define r4 (dst + 4 * n)
36#define v0 r0
37#define v1 r2
38#define vinf r4
39
40 // v2=(v2-vm1)/3
41 if (vm1_neg)
42 lmmp_add_n_(v2, v2, vm1, dnp);
43 else
44 lmmp_sub_n_(v2, v2, vm1, dnp);
46
47 // vm1=(v1-vm1)/2
48 if (vm1_neg)
49 lmmp_shr1add_n_(vm1, v1, vm1, dnp);
50 else
51 lmmp_shr1sub_n_(vm1, v1, vm1, dnp);
52
53 // v1=v1-v0
54 v1[2 * n] -= lmmp_sub_n_(v1, v1, v0, 2 * n);
55
56 // v2=(v2-v1)/2
57 lmmp_shr1sub_n_(v2, v2, v1, dnp);
58
59 // v1=v1-vm1
60 lmmp_sub_n_(v1, v1, vm1, dnp);
61
62 // add vm1 at correct place.
63 cy = lmmp_add_n_(r1, r1, vm1, dnp);
64 lmmp_inc_1(r3 + 1, cy); // at most propagate to v1[2*n]
65
66 saved = v1[2 * n]; // it is vinf[0]
67 vinf[0] = vinf0; // correct value of vinf
68
69 // v2=v2-vinf*2
70 cy = lmmp_shl_(vm1, vinf, spt, 1);
71 cy += lmmp_sub_n_(v2, v2, vm1, spt);
72 lmmp_dec_1(v2 + spt, cy);
73
74 // vinf+=v2h, no overflow
75 cy = lmmp_add_n_(vinf, vinf, v2 + n, n + 1);
76 lmmp_inc_1(r3 + dnp, cy);
77
78 // v1-=vinf, (same time vmh-=v2h)
79 cy = lmmp_sub_n_(v1, v1, vinf, spt);
80 vinf0 = vinf[0];
81 v1[2 * n] = saved; // correct value of v1
82 lmmp_dec_1(v1 + spt, cy);
83
84 // vml-=v2l
85 cy = lmmp_sub_n_(r1, r1, v2, n);
86 lmmp_dec_1(v1, cy);
87
88 // last v2l
89 cy = lmmp_add_n_(r3, r3, v2, n);
90 v1[2 * n] += cy; // no carry
91 lmmp_inc_1(vinf, vinf0);
92}
void lmmp_divexact_by3_(mp_ptr dst, mp_srcptr numa, mp_size_t na)
Copyright (C) 2026 HJimmyK (Jericho Knox)
mp_limb_t * mp_ptr
定义 lmmp.h:96
uint64_t mp_size_t
定义 lmmp.h:93
uint64_t mp_limb_t
定义 lmmp.h:92
mp_limb_t lmmp_shr1add_n_(mp_ptr dst, mp_srcptr numa, mp_srcptr numb, mp_size_t n)
加法后右移1位 [dst,n] = ([numa,n] + [numb,n]) >> 1
定义 shr.c:65
mp_limb_t lmmp_shl_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_size_t shl)
左移操作 [dst,na] = [numa,na]<<shl,dst的低shl位填充0
定义 shl.c:26
#define lmmp_dec_1(p, dec)
减指定值宏(预期无借位)
mp_limb_t lmmp_sub_n_(mp_ptr dst, mp_srcptr numa, mp_srcptr numb, mp_size_t n)
无借位的n位减法 [dst,n] = [numa,n] - [numb,n]
定义 sub_n.c:84
mp_limb_t lmmp_shr1sub_n_(mp_ptr dst, mp_srcptr numa, mp_srcptr numb, mp_size_t n)
减法后右移1位 [dst,n] = ([numa,n] - [numb,n]) >> 1
定义 shr.c:87
#define lmmp_inc_1(p, inc)
加指定值宏(预期无进位)
mp_limb_t lmmp_add_n_(mp_ptr dst, mp_srcptr numa, mp_srcptr numb, mp_size_t n)
无进位的n位加法 [dst,n] = [numa,n] + [numb,n]
定义 add_n.c:76
#define v0
#define vinf
#define vm1
#define r1
#define v1
#define r3
#define v2
#define n
void lmmp_toom_interp5_(mp_ptr dst, mp_ptr v2, mp_ptr vm1, mp_size_t n, mp_size_t spt, int vm1_neg, mp_limb_t vinf0)
Copyright (C) 2026 HJimmyK (Jericho Knox)