#pragma GCC optimization "Ofast"
#pragma GCC optimization "unroll-loop"
#pragma GCC target ("avx2")
#include <bits/stdc++.h>
#define ll int
#define ld long double
#define fs first
#define sc second
using namespace std;
const ll N = 3e4 + 6;
const ll Log2 = 19;
const ll inf = 1e9 + 7;
typedef pair<ll,ll> LL;
vector<LL> v;
struct doge{
ll pos,p,id;
};
doge a[N];
ll n,m,d[N];
map<LL,ll> mp;
vector<LL> g[N];
void Dij(){
fill(d,d + N - 1,inf);
priority_queue<LL,vector<LL>,greater<LL>> pq; pq.push({0,a[1].pos}); d[a[1].pos] = 0;
while(!pq.empty()){
LL t = pq.top(); pq.pop();
ll u = t.sc;
for (auto i : g[u]){
ll v = i.fs,L = i.sc;
if (d[v] > d[u] + L) d[v] = d[u] + L,pq.push({d[v],v});
}
}
//for (auto i : g[1]) cout<<i.fs<<"x "; exit(0);
//cout<<d[2]; exit(0);
if (d[a[2].pos] != inf) cout<<d[a[2].pos];
else cout<<-1;
}
int main(){
ios_base::sync_with_stdio(0);
cin.tie(0), cout.tie(0);
#define task "test"
if (fopen(task".inp", "r")){
freopen(task".inp", "r", stdin);
//freopen(task".out", "w", stdout);
}
cin>>n>>m;
for (ll i = 1;i <= m;i++){
cin>>a[i].pos>>a[i].p,a[i].pos++;
LL now = {a[i].pos,a[i].p};
mp[now] = 1;
}
for (ll i = 1;i <= m;i++){
LL now = {a[i].pos,a[i].p};
ll cost = 1;
for (ll j = now.fs + a[i].p;j <= n;j += a[i].p){
//if (mp[{j,now.sc}]) break;
//mp[{j,now.sc}] = 1; //cout<<j<<" ";
g[now.fs].push_back({j,cost}),cost++;
if (mp[{j,now.sc}]) break;
}
cost = 1; //return 0;
for (ll j = now.fs - a[i].p;j > 0;j -= a[i].p){
//
//mp[{j,now.sc}] = 1;
g[now.fs].push_back({j,cost}),cost++;
if (mp[{j,now.sc}]) break;
//if (now.fs == 5) cout<<j<<" x ";
}
}
Dij();
}
