代码

#include <stdio.h>
#include <stdlib.h>
#include <immintrin.h>

#pragma GCC target("AVX2")
void solve(int n, int q, char *s1, char *s2, int *q_x, int *q_y, int *q_len, unsigned *ans) {
    int *a1 = (int*)malloc(n * sizeof(int));
    int *a2 = (int*)malloc(n * sizeof(int));
    for (int i = 0; i < n; ++i) {
        a1[i] = s1[i] - '0';
        a2[i] = s2[i] - '0';
    }

    for (int i = 0; i < q; ++i) {
        int x = q_x[i];
        int y = q_y[i];
        int l = q_len[i];
        unsigned sum = 0;
        int *p1 = a1 + x;
        int *p2 = a2 + y;

        int k = 0;
        const int vec_size = 8; // AVX2处理8个int32
        int limit = l - (l % vec_size);
        __m256i sum_v = _mm256_setzero_si256();
        __m256i three = _mm256_set1_epi32(3);
        for (; k < limit; k += vec_size) {
            __m256i va = _mm256_loadu_si256((__m256i*)(p1 + k));
            __m256i vb = _mm256_loadu_si256((__m256i*)(p2 + k));
            __m256i diff = _mm256_sub_epi32(va, vb);
            __m256i v_plus3 = _mm256_add_epi32(diff, three);

            // 检查v_plus3是否等于2或5
            __m256i cmp2 = _mm256_cmpeq_epi32(v_plus3, _mm256_set1_epi32(2));
            __m256i cmp5 = _mm256_cmpeq_epi32(v_plus3, _mm256_set1_epi32(5));
            __m256i or_result = _mm256_or_si256(cmp2, cmp5);

            // 将比较结果转换为1或0
            or_result = _mm256_srai_epi32(or_result, 31); // 结果为全1或全0
            or_result = _mm256_and_si256(or_result, _mm256_set1_epi32(1));

            sum_v = _mm256_add_epi32(sum_v, or_result);
        }

        // 累加sum_v中的结果
        int temp[8] __attribute__((aligned(32)));
        _mm256_store_si256((__m256i*)temp, sum_v);
        for (int j = 0; j < 8; ++j) {
            sum += temp[j];
        }

        // 处理剩余元素
        for (; k < l; ++k) {
            int a = p1[k];
            int b = p2[k];
            sum += ((a - b + 3) % 3) == 2;
        }

        ans[i] = sum;
    }

    free(a1);
    free(a2);
}

评测结果