LMMP 4.3.0
Lamina High-Precision Arithmetic Library
载入中...
搜索中...
未找到
mul_basecase.c 文件参考
#include "impl/longlong.h"
#include "lmmp/lmmpn.h"
mul_basecase.c 的引用(Include)关系图:

浏览该文件的源代码.

函数

void lmmp_mul_basecase_ (mp_ptr restrict dst, mp_srcptr restrict numa, mp_size_t na, mp_srcptr restrict numb, mp_size_t nb)
void lmmp_sqr_basecase_ (mp_ptr restrict dst, mp_srcptr restrict numa, mp_size_t na)
 Copyright (C) 2026 HJimmyK (Jericho Knox)

函数说明

◆ lmmp_mul_basecase_()

void lmmp_mul_basecase_ ( mp_ptr restrict dst,
mp_srcptr restrict numa,
mp_size_t na,
mp_srcptr restrict numb,
mp_size_t nb )

在文件 mul_basecase.c 第 106 行定义.

112 {
113 lmmp_param_assert(na >= nb);
114 lmmp_param_assert(nb >= 1);
115
116 mp_size_t i;
117 mp_limb_t cl, x;
118
119 /* 首列:dst[0..na] = numb[0] * numa */
120 x = numb[0];
121 cl = 0;
122 for (i = 0; i + 4 <= na; i += 4) {
123 __uint128_t a0 = (__uint128_t)numa[i + 0] * x + cl;
124 __uint128_t a1 = (__uint128_t)numa[i + 1] * x + (uint64_t)(a0 >> 64);
125 __uint128_t a2 = (__uint128_t)numa[i + 2] * x + (uint64_t)(a1 >> 64);
126 __uint128_t a3 = (__uint128_t)numa[i + 3] * x + (uint64_t)(a2 >> 64);
127 dst[i + 0] = (uint64_t)a0;
128 dst[i + 1] = (uint64_t)a1;
129 dst[i + 2] = (uint64_t)a2;
130 dst[i + 3] = (uint64_t)a3;
131 cl = (uint64_t)(a3 >> 64);
132 }
133 for (; i < na; i++) {
134 __uint128_t a = (__uint128_t)numa[i] * x + cl;
135 dst[i] = (uint64_t)a;
136 cl = (uint64_t)(a >> 64);
137 }
138 dst[na] = cl;
139
140 /* 其余列:逐列 addmul 累加,两列合并外层循环 */
141 mp_size_t j = 1;
142 mp_ptr dp = dst + 1;
143 nb -= 1;
144 for (; nb >= 2; j += 2, dp += 2, nb -= 2) {
145 cl = 0;
146 x = numb[j];
147 for (i = 0; i + 4 <= na; i += 4) {
148 __uint128_t a0 = (__uint128_t)numa[i + 0] * x + dp[i + 0] + cl;
149 __uint128_t a1 = (__uint128_t)numa[i + 1] * x + dp[i + 1] + (uint64_t)(a0 >> 64);
150 __uint128_t a2 = (__uint128_t)numa[i + 2] * x + dp[i + 2] + (uint64_t)(a1 >> 64);
151 __uint128_t a3 = (__uint128_t)numa[i + 3] * x + dp[i + 3] + (uint64_t)(a2 >> 64);
152 dp[i + 0] = (uint64_t)a0;
153 dp[i + 1] = (uint64_t)a1;
154 dp[i + 2] = (uint64_t)a2;
155 dp[i + 3] = (uint64_t)a3;
156 cl = (uint64_t)(a3 >> 64);
157 }
158 for (; i < na; i++) {
159 __uint128_t a = (__uint128_t)numa[i] * x + dp[i] + cl;
160 dp[i] = (uint64_t)a;
161 cl = (uint64_t)(a >> 64);
162 }
163 dp[na] = cl;
164
165 cl = 0;
166 x = numb[j + 1];
167 for (i = 0; i + 4 <= na; i += 4) {
168 __uint128_t a0 = (__uint128_t)numa[i + 0] * x + dp[i + 1] + cl;
169 __uint128_t a1 = (__uint128_t)numa[i + 1] * x + dp[i + 2] + (uint64_t)(a0 >> 64);
170 __uint128_t a2 = (__uint128_t)numa[i + 2] * x + dp[i + 3] + (uint64_t)(a1 >> 64);
171 __uint128_t a3 = (__uint128_t)numa[i + 3] * x + dp[i + 4] + (uint64_t)(a2 >> 64);
172 dp[i + 1] = (uint64_t)a0;
173 dp[i + 2] = (uint64_t)a1;
174 dp[i + 3] = (uint64_t)a2;
175 dp[i + 4] = (uint64_t)a3;
176 cl = (uint64_t)(a3 >> 64);
177 }
178 for (; i < na; i++) {
179 __uint128_t a = (__uint128_t)numa[i] * x + dp[i + 1] + cl;
180 dp[i + 1] = (uint64_t)a;
181 cl = (uint64_t)(a >> 64);
182 }
183 dp[na + 1] = cl;
184 }
185 for (; nb >= 1; j++, dp++, nb--) {
186 cl = 0;
187 x = numb[j];
188 for (i = 0; i + 4 <= na; i += 4) {
189 __uint128_t a0 = (__uint128_t)numa[i + 0] * x + dp[i + 0] + cl;
190 __uint128_t a1 = (__uint128_t)numa[i + 1] * x + dp[i + 1] + (uint64_t)(a0 >> 64);
191 __uint128_t a2 = (__uint128_t)numa[i + 2] * x + dp[i + 2] + (uint64_t)(a1 >> 64);
192 __uint128_t a3 = (__uint128_t)numa[i + 3] * x + dp[i + 3] + (uint64_t)(a2 >> 64);
193 dp[i + 0] = (uint64_t)a0;
194 dp[i + 1] = (uint64_t)a1;
195 dp[i + 2] = (uint64_t)a2;
196 dp[i + 3] = (uint64_t)a3;
197 cl = (uint64_t)(a3 >> 64);
198 }
199 for (; i < na; i++) {
200 __uint128_t a = (__uint128_t)numa[i] * x + dp[i] + cl;
201 dp[i] = (uint64_t)a;
202 cl = (uint64_t)(a >> 64);
203 }
204 dp[na] = cl;
205 }
206}
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
#define lmmp_param_assert(x)
定义 lmmp.h:509
#define a0
#define a1
#define a2
#define a3
#define numb

引用了 a0, a1, a2, a3, lmmp_mul_basecase_(), lmmp_param_assert, numb , 以及 restrict.

函数调用图:

◆ lmmp_sqr_basecase_()

void lmmp_sqr_basecase_ ( mp_ptr restrict dst,
mp_srcptr restrict numa,
mp_size_t na )

Copyright (C) 2026 HJimmyK (Jericho Knox)

This file is part of LMMP.

LMMP is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version.

This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with this library. If not, see https://www.gnu.org/licenses/.

SPDX-License-Identifier: LGPL-3.0-or-later

在文件 mul_basecase.c 第 27 行定义.

27 {
28 lmmp_param_assert(na >= 1);
29
30 mp_size_t i, j;
31 mp_limb_t cl, x;
32
33 dst[0] = 0;
34 x = numa[0];
35 cl = 0;
36 for (i = 1; i + 4 <= na; i += 4) {
37 __uint128_t a0 = (__uint128_t)numa[i] * x + cl;
38 __uint128_t a1 = (__uint128_t)numa[i + 1] * x + (mp_limb_t)(a0 >> 64);
39 __uint128_t a2 = (__uint128_t)numa[i + 2] * x + (mp_limb_t)(a1 >> 64);
40 __uint128_t a3 = (__uint128_t)numa[i + 3] * x + (mp_limb_t)(a2 >> 64);
41 dst[i] = (mp_limb_t)a0;
42 dst[i + 1] = (mp_limb_t)a1;
43 dst[i + 2] = (mp_limb_t)a2;
44 dst[i + 3] = (mp_limb_t)a3;
45 cl = (mp_limb_t)(a3 >> 64);
46 }
47 for (; i < na; i++) {
48 __uint128_t a = (__uint128_t)numa[i] * x + cl;
49 dst[i] = (mp_limb_t)a;
50 cl = (mp_limb_t)(a >> 64);
51 }
52 dst[na] = cl;
53
54 for (j = 1; j + 1 < na; j++) {
55 x = numa[j];
56 cl = 0;
57 for (i = j + 1; i + 4 <= na; i += 4) {
58 __uint128_t a0 = (__uint128_t)numa[i] * x + dst[i + j] + cl;
59 __uint128_t a1 = (__uint128_t)numa[i + 1] * x + dst[i + j + 1] + (mp_limb_t)(a0 >> 64);
60 __uint128_t a2 = (__uint128_t)numa[i + 2] * x + dst[i + j + 2] + (mp_limb_t)(a1 >> 64);
61 __uint128_t a3 = (__uint128_t)numa[i + 3] * x + dst[i + j + 3] + (mp_limb_t)(a2 >> 64);
62 dst[i + j] = (mp_limb_t)a0;
63 dst[i + j + 1] = (mp_limb_t)a1;
64 dst[i + j + 2] = (mp_limb_t)a2;
65 dst[i + j + 3] = (mp_limb_t)a3;
66 cl = (mp_limb_t)(a3 >> 64);
67 }
68 for (; i < na; i++) {
69 __uint128_t a = (__uint128_t)numa[i] * x + dst[i + j] + cl;
70 dst[i + j] = (mp_limb_t)a;
71 cl = (mp_limb_t)(a >> 64);
72 }
73 dst[na + j] = cl;
74 }
75
76 cl = 0;
77 for (i = 1; i + 4 <= 2 * na - 1; i += 4) {
78 __uint128_t a0 = (__uint128_t)dst[i] * 2 + cl;
79 __uint128_t a1 = (__uint128_t)dst[i + 1] * 2 + (mp_limb_t)(a0 >> 64);
80 __uint128_t a2 = (__uint128_t)dst[i + 2] * 2 + (mp_limb_t)(a1 >> 64);
81 __uint128_t a3 = (__uint128_t)dst[i + 3] * 2 + (mp_limb_t)(a2 >> 64);
82 dst[i] = (mp_limb_t)a0;
83 dst[i + 1] = (mp_limb_t)a1;
84 dst[i + 2] = (mp_limb_t)a2;
85 dst[i + 3] = (mp_limb_t)a3;
86 cl = (mp_limb_t)(a3 >> 64);
87 }
88 for (; i < 2 * na - 1; i++) {
89 __uint128_t a = (__uint128_t)dst[i] * 2 + cl;
90 dst[i] = (mp_limb_t)a;
91 cl = (mp_limb_t)(a >> 64);
92 }
93 dst[2 * na - 1] = cl;
94
95 cl = 0;
96 for (i = 0; i < na; i++) {
97 __uint128_t t = (__uint128_t)numa[i] * numa[i] + dst[2 * i] + cl;
98 dst[2 * i] = (mp_limb_t)t;
99 cl = (mp_limb_t)(t >> 64);
100 t = (__uint128_t)dst[2 * i + 1] + cl;
101 dst[2 * i + 1] = (mp_limb_t)t;
102 cl = (mp_limb_t)(t >> 64);
103 }
104}
#define t

引用了 a0, a1, a2, a3, lmmp_param_assert, lmmp_sqr_basecase_(), restrict , 以及 t.

函数调用图: