112 {
115
118
119
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
143 nb -= 1;
144 for (; nb >= 2; j += 2, dp += 2, nb -= 2) {
145 cl = 0;
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;
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;
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}
#define lmmp_param_assert(x)