#ifndef LOCAL
#define FAST_IO
#endif

// ============
#include <bits/stdc++.h>
#define OVERRIDE(a, b, c, d, ...) d
#define REP2(i, n) for (i32 i = 0; i < (i32)(n); ++i)
#define REP3(i, m, n) for (i32 i = (i32)(m); i < (i32)(n); ++i)
#define REP(...) OVERRIDE(__VA_ARGS__, REP3, REP2)(__VA_ARGS__)
#define PER2(i, n) for (i32 i = (i32)(n)-1; i >= 0; --i)
#define PER3(i, m, n) for (i32 i = (i32)(n)-1; i >= (i32)(m); --i)
#define PER(...) OVERRIDE(__VA_ARGS__, PER3, PER2)(__VA_ARGS__)
#define ALL(x) begin(x), end(x)
#define LEN(x) (i32)(x.size())
using namespace std;
using u32 = unsigned int;
using u64 = unsigned long long;
using i32 = signed int;
using i64 = signed long long;
using f64 = double;
using f80 = long double;
using pi = pair<i32, i32>;
using pl = pair<i64, i64>;
template <typename T>
using V = vector<T>;
template <typename T>
using VV = V<V<T>>;
template <typename T>
using VVV = V<V<V<T>>>;
template <typename T>
using VVVV = V<V<V<V<T>>>>;
template <typename T>
using PQR = priority_queue<T, V<T>, greater<T>>;
template <typename T>
bool chmin(T &x, const T &y) {
    if (x > y) {
        x = y;
        return true;
    }
    return false;
}
template <typename T>
bool chmax(T &x, const T &y) {
    if (x < y) {
        x = y;
        return true;
    }
    return false;
}
template <typename T>
i32 lob(const V<T> &arr, const T &v) {
    return (i32)(lower_bound(ALL(arr), v) - arr.begin());
}
template <typename T>
i32 upb(const V<T> &arr, const T &v) {
    return (i32)(upper_bound(ALL(arr), v) - arr.begin());
}
template <typename T>
V<i32> argsort(const V<T> &arr) {
    V<i32> ret(arr.size());
    iota(ALL(ret), 0);
    sort(ALL(ret), [&](i32 i, i32 j) -> bool {
        if (arr[i] == arr[j]) {
            return i < j;
        } else {
            return arr[i] < arr[j];
        }
    });
    return ret;
}
#ifdef INT128
using u128 = __uint128_t;
using i128 = __int128_t;
#endif
[[maybe_unused]] constexpr i32 INF = 1000000100;
[[maybe_unused]] constexpr i64 INF64 = 3000000000000000100;
struct SetUpIO {
    SetUpIO() {
#ifdef FAST_IO
        ios::sync_with_stdio(false);
        cin.tie(nullptr);
#endif
        cout << fixed << setprecision(15);
    }
} set_up_io;
void scan(char &x) { cin >> x; }
void scan(u32 &x) { cin >> x; }
void scan(u64 &x) { cin >> x; }
void scan(i32 &x) { cin >> x; }
void scan(i64 &x) { cin >> x; }
void scan(string &x) { cin >> x; }
template <typename T>
void scan(V<T> &x) {
    for (T &ele : x) {
        scan(ele);
    }
}
void read() {}
template <typename Head, typename... Tail>
void read(Head &head, Tail &...tail) {
    scan(head);
    read(tail...);
}
#define CHAR(...)     \
    char __VA_ARGS__; \
    read(__VA_ARGS__);
#define U32(...)     \
    u32 __VA_ARGS__; \
    read(__VA_ARGS__);
#define U64(...)     \
    u64 __VA_ARGS__; \
    read(__VA_ARGS__);
#define I32(...)     \
    i32 __VA_ARGS__; \
    read(__VA_ARGS__);
#define I64(...)     \
    i64 __VA_ARGS__; \
    read(__VA_ARGS__);
#define STR(...)        \
    string __VA_ARGS__; \
    read(__VA_ARGS__);
#define VEC(type, name, size) \
    V<type> name(size);       \
    read(name);
#define VVEC(type, name, size1, size2)    \
    VV<type> name(size1, V<type>(size2)); \
    read(name);
// ============

#ifdef DEBUGF
#else
#define DBG(...) (void)0
#endif

#include <atcoder/modint>
#include <atcoder/convolution>

// ============

#include <vector>
#include <cassert>

template <typename T>
class FactorialTable {
    std::vector<T> fac;
    std::vector<T> ifac;
    
public:
    FactorialTable() : fac(1, T(1)), ifac(1, T(1)) {}
    
    FactorialTable(int n) : fac(n + 1), ifac(n + 1) {
        assert(n >= 0);
        fac[0] = T(1);
        for (int i = 1; i <= n; ++i) {
            fac[i] = fac[i - 1] * T(i);
        }
        ifac[n] = T(1) / T(fac[n]);
        for (int i = n; i > 0; --i) {
            ifac[i - 1] = ifac[i] * T(i);
        }
    }
    
    void resize(int n) {
        int old = n_max();
        if (n <= old) {
            return;
        }
        fac.resize(n + 1);
        for (int i = old + 1; i <= n; ++i) {
            fac[i] = fac[i - 1] * T(i);
        }
        ifac.resize(n + 1);
        ifac[n] = T(1) / T(fac[n]);
        for (int i = n; i > old; --i) {
            ifac[i - 1] = ifac[i] * T(i);
        }
    }
    
    inline int n_max() const {
        return (int) fac.size() - 1;
    }
    
    inline T fact(int n) const {
        assert(n >= 0 && n <= n_max());
        return fac[n];
    }
    
    inline T inv_fact(int n) const {
        assert(n >= 0 && n <= n_max());
        return ifac[n];
    }
    
    inline T inv_n(int n) const {
        assert(n > 0 && n <= n_max());
        return ifac[n] * fac[n - 1];
    }
    
    inline T choose(int n, int k) const {
        assert(k <= n_max() && n <= n_max());
        if (k > n || k < 0) {
            return T(0);
        }
        return fac[n] * ifac[k] * ifac[n - k];
    }
    
    inline T multi_choose(int n, int k) const {
        assert(n >= 1 && k >= 0 && k + n - 1 <= n_max());
        return choose(k + n - 1, k);
    }
    
    inline T n_terms_sum_k(int n, int k) const {
        assert(n >= 0);
        if (k < 0) {
            return T(0);
        }
        if (n == 0) {
            return k == 0 ? T(1) : T(0);
        }
        return choose(n + k - 1, n - 1);
    }
};
// ============

using DM = atcoder::dynamic_modint<998>;
using M9 = atcoder::modint998244353;

void solve() {
    I64(n);
    I32(p);
    DM::set_mod(p);
    
    auto check_order = [&](i32 x) -> i32 {
        DM val(x);
        i32 e = 1;
        while (val.val() != 1) {
            val *= DM(x);
            ++e;
        }
        return e;
    };
    
    i32 root = 1;
    while (check_order(root) != p - 1) {
        ++root;
    }
    
    V<i32> ord(p, 0);
    {
        DM pw(1);
        REP(i, p - 1) {
            ord[pw.val()] = i;
            pw *= DM(root);
        }
    }
    DBG(root, ord);
    
    FactorialTable<DM> fac(p - 1);
    
    V<i32> ds;
    while (n > 0) {
        ds.push_back(n % p);
        n /= p;
    }
    
    V<M9> dp(p - 1);
    dp[0] = M9(1);
    for (i32 d : ds) {
        V<M9> tmp(p - 1);
        REP(i, d + 1) {
            DM binom = fac.choose(d, i);
            tmp[ord[binom.val()]] += M9(1);
        }
        V<M9> conv = atcoder::convolution(dp, tmp);
        dp.assign(p - 1, M9());
        REP(i, LEN(conv)) {
            dp[i % (p - 1)] += conv[i];
        }
    }
    M9 ans;
    DM pw(1);
    REP(i, p - 1) {
        ans += M9(pw.val()) * dp[i];
        pw *= DM(root);
    }
    cout << ans.val() << '\n';
}

int main() {
    i32 t = 1;
    // cin >> t;
    while (t--) {
        solve();
    }
}