Different optimizations in VS2015 and VS2013 cause a floating point exception

Question

I have a small example of the problem that occurred when switching from VS2013 to VS2015. In VS2015, the following code example causes an invalid floating point operation.

int main()
{
    unsigned int enableBits = _EM_OVERFLOW | _EM_ZERODIVIDE | _EM_INVALID;

    _clearfp();
    _controlfp_s(0, ~enableBits, enableBits);

    int count = 100;
    float array[100];

    for (int i = 0; i < count; ++i)
    {
        array[i] = (float)pow((float)(count - 1 - i) / count, 4); //this causes exception in VS2015
    }

    return 0;
}

This only happens in release mode, so it’s probably due to another optimization. Is there something wrong with this code or is this a bug in VS 2015?

It is difficult to find such problems on the entire code base, so I'm looking for some kind of systematic fix, and not a workaround (for example, use another variable instead of i that works)

, , , VS2013 128- 4 . VS2015, , 2 , ( ), , , .

, , .

VS2013

VS2015

. .

Answer 1

, .

, 2 ( ), divps, 4 ( 4 XMM). 2 XMM . , . , , op, , , .

- , , /fp: strict, , ( , , ) controlfp.