Trying to learn c++. Why does this seg fault at 9 million but

>>60031076
There is a reason interviewers only want you to sum the first two million primes.

i dont give a fuck about archaic shit like c(++) but maybe check ur types

Stack overflow perhaps? Try allocating that bitmap on the heap.

9million 32bit ints is 36megs so maybe a stackoverflow

>>60031076
> bool primes[++n];

>>60031123
>>60031143

Cool that got it.

>>60031160
Not correct? If you submit 20 it will only loop up to 19 with standard a < b syntax.

Creating the array with arr[++n] is correct then?

>>60031160
>>60031177
It creates the array of a fixed size?
Its too large an array

>>60031189
Actually I have no idea. The program works fine on random online compiler I pasted the code to.

Ok thanks everybody. Final question due to the lengths and the fact I am using a 64bit processor, should I use longs or ints?

#include <iostream>
#include <math.h>


long prime(int n) {
    std::cout << "starting\n";
    bool *primes = new bool[++n];

    for (long i = 0; i < sqrt(n) + 1; i++) {
        primes[i] = false;
    }

    for (long p = 2; p < n; p++) {
        if (!primes[p])
            for (long j = p * 2; j < n; j += p)
                primes[j] = true;
    }

    long sum = 0;
    for (int i = 0; i < n; i++)
        if (!primes[i])
            sum += i;

    return sum;
}

int main() {
    long primes = prime(2e6);
    std::cout << primes;
    return 0;
}

>>60031217
Memleak, you noob. If you can't handle manual memory - use smart pointers.

auto primes = std::make_unique(new bool[++n]);

>>60031217

Use unsigned long, since unsigned int only goes up to 2^16 - 1 obviously. Has nothing to do with you having a 64bit processor.

>>60031217
I don't know about the C++ standard but the size of data types isn't really guaranteed. You should be using the data types from inttypes.h.
eg. uint64_t, uint32_t, int8_t