代码

#pragma GCC optimize("Ofast")
#pragma GCC target("sse,sse2,sse3,ssse3,sse4")
int c0[2050], c1[2050], c2[2050];
unsigned bi[100000005];

void sort(unsigned *a, int n)
{
    unsigned *ai = a;
    for (register int i = 0; i < n; ++ i)
    {
        c0[ai[i] & 2047] ++;
        c1[(ai[i] >> 11) & 2047] ++;
        c2[ai[i] >> 22] ++;
    }
    for (int i = 1; i < (1 << 11); ++ i) c0[i] += c0[i - 1];
    for (int i = 1; i < (1 << 11); ++ i) c1[i] += c1[i - 1];
    for (int i = 1; i < (1 << 11); ++ i) c2[i] += c2[i - 1];

    for (register int i = n - 1; ~i; -- i) bi[-- c0[ai[i] & ((1 << 11) - 1)]] = ai[i];
    for (register int i = n - 1; ~i; -- i) ai[-- c1[(bi[i] >> 11) & ((1 << 11) - 1)]] = bi[i];
    for (register int i = n - 1; ~i; -- i) bi[-- c2[ai[i] >> 22]] = ai[i];

    for (register int i = 0; i < n; ++ i) ai[i] = bi[i];
}

评测结果