代码

// fileName: test_O2_Unroll_loops_compiler.cpp
#pragma GCC optimize("O2,unroll-loops")

#include <cstdio>
#include <random>
#include <chrono>

std::mt19937 rng(std::chrono::steady_clock::now().time_since_epoch().count());

typedef unsigned int UI;
typedef unsigned long long ULL;

const UI Mod = 998244353;
const int N = 1400;

UI A[N][N], B[N][N], C[N][N];

void Mul() {
	for (int i = 0; i < N; ++i)
		for (int j = 0; j < N; ++j)
			for (int k = 0; k < N; ++k)
				C[i][j] = (C[i][j] + (ULL)A[i][k] * B[k][j]) % Mod;
}

#ifdef PKRBT
int main() {
#else
void matrix_multiply(int, const double *, const double *, double *) {
#endif
	for (int i = 0; i < N; ++i) {
		for (int j = 0; j < N; ++j) {
			A[i][j] = rng() % Mod;
			B[i][j] = rng() % Mod;
			C[i][j] = 0;
		}
	}
	Mul();
	UI Ans = 0;
	for (int i = 0; i < N; ++i)
		for (int j = 0; j < N; ++j)
			Ans ^= C[i][j];
	printf("%u\n", Ans);
#ifdef PKRBT
	return 0;
#endif
}

评测结果