#include<bits/stdc++.h>
#include<atcoder/all>
#define rep(i,b) for(int i=0;i<b;i++)
#define rrep(i,b) for(int i=b-1;i>=0;i--)
#define rep1(i,b) for(int i=1;i<b;i++)
#define repx(i,x,b) for(int i=x;i<b;i++)
#define fore(i,a) for(auto i:a)
#define fore1(i,a) for(auto &i:a)
#define rng(x) (x).begin(), (x).end()
#define rrng(x) (x).rbegin(), (x).rend()
#define sz(x) ((int)(x).size())
#define pb push_back
#define fi first
#define se second

using namespace std;
using namespace atcoder;

using ll = long long;
template<typename T> using mpq = priority_queue<T, vector<T>, greater<T>>;
template<typename T> bool chmax(T &a, const T &b) { if (a<b) { a=b; return 1; } return 0; }
template<typename T> bool chmin(T &a, const T &b) { if (b<a) { a=b; return 1; } return 0; }
template<typename T> ll sumv(const vector<T>&a){ll res(0);for(auto&&x:a)res+=x;return res;}
bool yn(bool a) { if(a) {cout << "Yes" << endl; return 1;} else {cout << "No" << endl; return 0;}}
#define dame { cout << "No" << endl; return 0;}
#define dame1 { cout << -1 << endl; return 0;}
#define test(x) cout << "test" << x << endl;
#define deb(x,y) cout << x << " " << y << endl;
#define deb3(x,y,z) cout << x << " " << y << " " << z << endl;
#define deb4(x,y,z,x2) cout << x << " " << y << " " << z << " " << x2 << endl;
#define out cout << ans << endl;
#define outv fore(yans , ans) cout << yans << "\n";
#define show(x) cerr<<#x<<" = "<<x<<endl;
#define showv(v) {fore(vy , v) {cout << vy << " ";} cout << endl;}
#define showv2(v) fore(vy , v) cout << vy << "\n";
using pll = pair<ll,ll>;
using pil = pair<int,ll>;
using pli = pair<ll,int>;
using pii = pair<int,int>;
using tp = tuple<int ,int ,int>;
using vi = vector<int>;
using vl = vector<ll>;
using vs = vector<string>;
using vb = vector<bool>;
using vpii = vector<pii>;
using vpli = vector<pli>;
using vpll = vector<pll>;
using vpil = vector<pil>;
using vvi = vector<vector<int>>;
using vvl = vector<vector<ll>>;
using vvs = vector<vector<string>>;
using vvb = vector<vector<bool>>;
using vvpii = vector<vector<pii>>;
using vvpli = vector<vector<pli>>;
using vvpll = vector<vpll>;
using vvpil = vector<vpil>;
using mint = modint998244353;
using vm = vector<mint>;
using vvm = vector<vector<mint>>;
vector<int> dx={1,0,-1,0,1,1,-1,-1},dy={0,1,0,-1,1,-1,1,-1};
ll gcd(ll a, ll b) { return a?gcd(b%a,a):b;}
ll lcm(ll a, ll b) { return a/gcd(a,b)*b;}
const double eps = 1e-10;
const ll LINF = 1001002003004005006ll;
const int INF = 1001001001;

int main(){
    int n,m; cin>>n>>m;

    vector<fenwick_tree<mint>> fw(m , fenwick_tree<mint>(n+2));

    mint ans = 0;

    rep(i,n+1){

        vm sum(m);

        rep(j,m) sum[j] = fw[j].sum(0,i+1);

        // show(i);
        // rep(j,m) deb(j , sum[j].val());

        if (i == n){
            rep(j,m) ans += sum[j];
            break;
        }else if(i == 0){
            rep(j,m){
                int ni = i + j + 1;
                fw[j].add(i+1,1);
                fw[j].add(min(ni , n+1) , -1);
            }
            continue;
        }

        mint all = 0;

        rep(j,m) all += sum[j];

        rep(j,m){
            int ni = i + j + 1;
            fw[j].add(i+1,all - sum[j]);
            fw[j].add(min(ni , n+1) , -(all - sum[j]));
        }
    }

    ans = mint(m).pow(n) - ans;

    cout << ans.val() << endl;

    return 0;
}