#pragma GCC optimize ("Ofast")
#include<bits/stdc++.h>
using namespace std;
inline int my_getchar_unlocked(){
  static char buf[1048576];
  static int s = 1048576;
  static int e = 1048576;
  if(s == e && e == 1048576){
    e = fread_unlocked(buf, 1, 1048576, stdin);
    s = 0;
  }
  if(s == e){
    return EOF;
  }
  return buf[s++];
}
inline void rd(int &x){
  int k;
  int m=0;
  x=0;
  for(;;){
    k = my_getchar_unlocked();
    if(k=='-'){
      m=1;
      break;
    }
    if('0'<=k&&k<='9'){
      x=k-'0';
      break;
    }
  }
  for(;;){
    k = my_getchar_unlocked();
    if(k<'0'||k>'9'){
      break;
    }
    x=x*10+k-'0';
  }
  if(m){
    x=-x;
  }
}
struct MY_WRITER{
  char buf[1048576];
  int s;
  int e;
  MY_WRITER(){
    s = 0;
    e = 1048576;
  }
  ~MY_WRITER(){
    if(s){
      fwrite_unlocked(buf, 1, s, stdout);
    }
  }
}
;
MY_WRITER MY_WRITER_VAR;
void my_putchar_unlocked(int a){
  if(MY_WRITER_VAR.s == MY_WRITER_VAR.e){
    fwrite_unlocked(MY_WRITER_VAR.buf, 1, MY_WRITER_VAR.s, stdout);
    MY_WRITER_VAR.s = 0;
  }
  MY_WRITER_VAR.buf[MY_WRITER_VAR.s++] = a;
}
inline void wt_L(char a){
  my_putchar_unlocked(a);
}
inline void wt_L(int x){
  int s=0;
  int m=0;
  char f[10];
  if(x<0){
    m=1;
    x=-x;
  }
  while(x){
    f[s++]=x%10;
    x/=10;
  }
  if(!s){
    f[s++]=0;
  }
  if(m){
    my_putchar_unlocked('-');
  }
  while(s--){
    my_putchar_unlocked(f[s]+'0');
  }
}
template<class S, class T> inline S chmax(S &a, T b){
  if(a<b){
    a=b;
  }
  return a;
}
int N;
int A[16];
int B[16];
int C[16];
int dp[65536][16][3];
void gt(int j, int k, int &x, int &y){
  if(k==0){
    x = B[j];
    y = C[j];
  }
  if(k==1){
    x = A[j];
    y = C[j];
  }
  if(k==2){
    x = A[j];
    y = B[j];
  }
  if(x > y){
    swap(x, y);
  }
  ;
}
int main(){
  int i, mask;
  int res = 0;
  int bx;
  int by;
  int nx;
  int ny;
  rd(N);
  {
    int Lj4PdHRW;
    for(Lj4PdHRW=(0);Lj4PdHRW<(N);Lj4PdHRW++){
      rd(A[Lj4PdHRW]);
      rd(B[Lj4PdHRW]);
      rd(C[Lj4PdHRW]);
    }
  }
  for(i=(0);i<(N);i++){
    dp[1<<i][i][0] = A[i];
    dp[1<<i][i][1] = B[i];
    dp[1<<i][i][2] = C[i];
  }
  for(mask=(0);mask<(1<<N);mask++){
    int j;
    for(j=(0);j<(N);j++){
      int k;
      for(k=(0);k<(3);k++){
        if(dp[mask][j][k]){
          int nj;
          chmax(res, dp[mask][j][k]);
          gt(j,k,bx,by);
          for(nj=(0);nj<(N);nj++){
            if(!(mask&1<<nj)){
              int nk;
              for(nk=(0);nk<(3);nk++){
                gt(nj,nk,nx,ny);
                if(nx > bx || ny > by){
                  continue;
                }
                chmax(dp[mask | (1<<nj)][nj][nk], dp[mask][j][k] + (A[nj] + B[nj] + C[nj] - nx - ny));
              }
            }
          }
        }
      }
    }
  }
  wt_L(res);
  wt_L('\n');
  return 0;
}
// cLay varsion 20200419-1

// --- original code ---
// int N, A[16], B[16], C[16];
// int dp[65536][16][3];
// 
// void gt(int j, int k, int &x, int &y){
//   if(k==0) x = B[j], y = C[j];
//   if(k==1) x = A[j], y = C[j];
//   if(k==2) x = A[j], y = B[j];
//   sortE(x, y);
// }
// 
// {
//   int res = 0, bx, by, nx, ny;
//   rd(N,(A,B,C)(N));
//   rep(i,N){
//     dp[1<<i][i][0] = A[i];
//     dp[1<<i][i][1] = B[i];
//     dp[1<<i][i][2] = C[i];
//   }
//   rep(mask,1<<N) rep(j,N) rep(k,3) if(dp[mask][j][k]){
//     res >?= dp[mask][j][k];
//     gt(j,k,bx,by);
//     rep(nj,N) if(!(mask&1<<nj)) rep(nk,3){
//       gt(nj,nk,nx,ny);
//       if(nx > bx || ny > by) continue;
//       dp[mask | (1<<nj)][nj][nk] >?= dp[mask][j][k] + (A[nj] + B[nj] + C[nj] - nx - ny);
//     }
//   }
// 
//   wt(res);
// }