#include <bits/stdc++.h>
#include <atcoder/all>

using namespace std;

/**
 * @brief Binary-Indexed-Tree(BIT)
 * @docs docs/binary-indexed-tree.md
 */
template< typename T >
struct BinaryIndexedTree {
private:
  int n;
  vector< T > data;

public:
  BinaryIndexedTree() = default;

  explicit BinaryIndexedTree(int n) : n(n) {
    data.assign(n + 1, 0);
  }

  explicit BinaryIndexedTree(const vector< T > &v) :
      BinaryIndexedTree((int) v.size()) {
    build(v);
  }

  void build(const vector< T > &v) {
    assert(n == (int) v.size());
    for(int i = 1; i <= n; i++) data[i] = v[i - 1];
    for(int i = 1; i <= n; i++) {
      int j = i + (i & -i);
      if(j <= n) data[j] += data[i];
    }
  }

  void apply(int k, const T &x) {
    for(++k; k <= n; k += k & -k) data[k] += x;
  }

  T prod(int r) const {
    T ret = T();
    for(; r > 0; r -= r & -r) ret += data[r];
    return ret;
  }

  T prod(int l, int r) const {
    return prod(r) - prod(l);
  }

  int lower_bound(T x) const {
    int i = 0;
    for(int k = 1 << (__lg(n) + 1); k > 0; k >>= 1) {
      if(i + k <= n && data[i + k] < x) {
        x -= data[i + k];
        i += k;
      }
    }
    return i;
  }

  int upper_bound(T x) const {
    int i = 0;
    for(int k = 1 << (__lg(n) + 1); k > 0; k >>= 1) {
      if(i + k <= n && data[i + k] <= x) {
        x -= data[i + k];
        i += k;
      }
    }
    return i;
  }
};

/**
 * @brief Abstract 2D Binary Indexed Tree Compressed(抽象化2次元座圧BIT)
 */
template< typename T >
struct Abstract2DBinaryIndexedTreeCompressed {
private:
  int n;
  vector< BinaryIndexedTree< T > > data;
  vector< int > hs;
  vector< vector< int > > beet;
public:
  Abstract2DBinaryIndexedTreeCompressed(const vector< int > &hs) :
      n((int) hs.size()), hs(hs) {
    vector< int > ord(n);
    iota(begin(ord), end(ord), 0);
    sort(begin(ord), end(ord), [&](int a, int b) {
      return hs[a] < hs[b];
    });
    beet.resize(n + 1);
    for(auto &&i: ord) {
      for(int k = i + 1; k <= n; k += k & -k) {
        beet[k].emplace_back(hs[i]);
      }
    }
    data.reserve(n + 1);
    for(int k = 0; k <= n; k++) {
      beet[k].erase(unique(begin(beet[k]), end(beet[k])), end(beet[k]));
      data.emplace_back(beet[k].size());
    }
  }

  void apply(int k1, const T &x) {
    int k2 = hs[k1];
    for(++k1; k1 <= n; k1 += k1 & -k1) {
      int p = lower_bound(begin(beet[k1]), end(beet[k1]), k2) - begin(beet[k1]);
      data[k1].apply(p, x);
    }
  }

  T prod(int r1, int r2) const {
    T ret{0};
    for(; r1 > 0; r1 -= r1 & -r1) {
      int p = lower_bound(begin(beet[r1]), end(beet[r1]), r2) - begin(beet[r1]);
      ret += data[r1].prod(p);
    }
    return ret;
  }
};

int main() {
  int M, K;
  cin >> M >> K;
  vector< int > A(M * K);
  for(auto &a: A) cin >> a;

  vector D(M, vector< int >());
  {
    for(int i = 0; i < M * K; i++) {
      D[A[i]].emplace_back(i);
      A[i] += 2 * (D[A[i]].size() - 1) * M;
    }
  }
  vector< int > vs(M * K * 2), ds(M * K * 2);
  BinaryIndexedTree< int > bit(M * K * 2);
  int64_t now = 0;
  for(int i = M * K - 1; i >= 0; i--) {
    now += bit.prod(A[i]);
    bit.apply(A[i], 1);
    vs[2 * i + 0] = A[i];
    ds[2 * i + 0] = 1;
    vs[2 * i + 1] = A[i] + M;
  }
  Abstract2DBinaryIndexedTreeCompressed< int > mat(vs);
  for(int i = 0; i < M * K; i++) {
    mat.apply(2 * i + 0, 1);
  }
  int64_t ret = now;
  for(int $ = 0; $ < M; $++) {
    for(auto &i: D[$]) {
      now -= i - mat.prod(2 * i, A[i]);
      now -= mat.prod(M * K * 2, A[i]) - mat.prod(2 * i + 2, A[i]);
      mat.apply(2 * i + 0, -1);
      mat.apply(2 * i + 1, 1);
      A[i] += M;
      now += i - mat.prod(2 * i, A[i]);
      now += mat.prod(M * K * 2, A[i]) - mat.prod(2 * i + 2, A[i]);
    }
    ret = min(ret, now);
  }
  cout << ret << "\n";
}