代码

//r1
#include <immintrin.h>
#include <string.h>
#include <cstdint>

#pragma GCC target("avx2")

inline void preprocess(char* s, int n) {
    for (int i = 0; i < n; ++i) {
        s[i] = '0' + ((s[i] - '0' + 2) % 3); // 正确映射
    }
}

static inline void process_blocks(int blocks, const char** p1, const char** p2, unsigned* sum) {
    for (int i = 0; i < blocks; ++i) {
        __m256i a1 = _mm256_loadu_si256((const __m256i*)(*p1));
        __m256i a2 = _mm256_loadu_si256((const __m256i*)(*p2));
        __m256i b1 = _mm256_loadu_si256((const __m256i*)(*p1 + 32));
        __m256i b2 = _mm256_loadu_si256((const __m256i*)(*p2 + 32));

        __m256i cmp1 = _mm256_cmpeq_epi8(a1, a2);
        __m256i cmp2 = _mm256_cmpeq_epi8(b1, b2);
        
        uint64_t mask = (uint64_t)_mm256_movemask_epi8(cmp2) << 32 
                      | _mm256_movemask_epi8(cmp1);
        *sum += __builtin_popcountll(mask);

        *p1 += 64;
        *p2 += 64;
    }
}

void solve(int n, int q, char* s1, char* s2, int* q_x, int* q_y, int* q_len, unsigned* ans) {
    preprocess(s1, n);

    for (int k = 0; k < q; ++k) {
        const char* p1 = s1 + q_x[k];
        const char* p2 = s2 + q_y[k];
        unsigned len = q_len[k];
        unsigned sum = 0;

        while (len >= 64) {
            int blocks = len / 64;
            if (blocks > 255) blocks = 255;
            process_blocks(blocks, &p1, &p2, &sum);
            len -= blocks * 64;
        }

        for (unsigned i = 0; i < len; ++i) {
            sum += (p1[i] == p2[i]);
        }

        ans[k] = sum;
    }
}

评测结果