#include <iostream>
#include <algorithm>
#include <iomanip>
#include <vector>
#include <queue>
#include <deque>
#include <set>
#include <map>
#include <tuple>
#include <cmath>
#include <numeric>
#include <functional>
#include <cassert>
#include <atcoder/modint>
#include <atcoder/convolution>

#define debug_value(x) cerr << "line" << __LINE__ << ":<" << __func__ << ">:" << #x << "=" << x << endl;
#define debug(x) cerr << "line" << __LINE__ << ":<" << __func__ << ">:" << x << endl;

template<class T> inline bool chmax(T& a, T b) { if (a < b) { a = b; return 1; } return 0; }
template<class T> inline bool chmin(T& a, T b) { if (a > b) { a = b; return 1; } return 0; }

using namespace std;
typedef long long ll;

template<typename T>
vector<vector<T>> vec2d(int n, int m, T v){
    return vector<vector<T>>(n, vector<T>(m, v));
}

template<typename T>
vector<vector<vector<T>>> vec3d(int n, int m, int k, T v){
    return vector<vector<vector<T>>>(n, vector<vector<T>>(m, vector<T>(k, v)));
}

template<typename T>
void print_vector(vector<T> v, char delimiter=' '){
    if(v.empty()) {
        cout << endl;
        return;
    }
    for(int i = 0; i+1 < v.size(); i++) cout << v[i] << delimiter;
    cout << v.back() << endl;
}

using mint = atcoder::modint998244353;

ostream& operator<<(ostream& os, const mint& m){
    os << m.val();
    return os;
}

#define N_MAX 200002
mint inv[N_MAX],fac[N_MAX],finv[N_MAX];

void init(){
    const ll MOD = mint::mod();
    fac[0]=1;fac[1]=1;
    finv[0]=1;finv[1]=1;
    inv[1]=1;
    for(int i=2;i<N_MAX;i++){
        inv[i]=mint(MOD)-inv[MOD%i]*(MOD/i);
        fac[i]=fac[i-1]*(ll) i;
        finv[i]=finv[i-1]*inv[i];
    }
}

mint comb(ll n, ll r){
    if(n < r){
        return mint(0);
    }else{
        return fac[n]*finv[r]*finv[n-r];
    }
}

vector<mint> f(int a, int b){
    if(a > b) swap(a, b);
    vector<mint> v(a+b+1);
    for(int x = 0; x <= a; x++){
        int len = a+b-x;
        v[len] = comb(b, x)*comb(a+b-x-1, b-1);
    }
    return v;
}

int main(){
    ios::sync_with_stdio(false);
    cin.tie(0);
    cout << setprecision(10) << fixed;
    init();
    int m, n; cin >> m >> n;
    vector<int> cnt(m);
    for(int i = 0; i < n; i++){
        int a; cin >> a; a--;
        cnt[a]++;
    }
    vector<int> cc(n+1);
    for(int i = 0; i < m; i++) cc[cnt[i]]++;
    vector<int> ccc;
    for(int i = 1; i <= n; i++){
        if(cc[i] >= 1) ccc.push_back(i);
    }
    vector<mint> f(n+1, 1);
    for(int i = 0; i <= n; i++){
        for(int x: ccc){
            if(comb(i, x).val() == 0) {
                f[i] = 0;
                continue;
            }
            f[i] *= comb(i, x).pow(cc[x]);
        }
    }
    vector<mint> v(n+1), u(n+1);
    for(int a = 0; a <= n; a++){
        v[a] = f[a]*(mint(-1).pow(a))*finv[a];
        u[a] = finv[a];
    }
    auto w = atcoder::convolution(u, v);
    mint ans = 0;
    for(int k = 0; k <= n; k++){
        ans += mint(-1).pow(k)*fac[k]*w[k];
    }
    for(int i = 0; i < m; i++) ans *= fac[cnt[i]];
    cout << ans << endl;
}