Ignore floating-point overflow and underflow errors in C++

Question

The following code snippet

void main()
{
   float a = 68440675640679078541805800652800.0f;
   float b = a*a;
   cout << b << endl;
}

produces a floating-point overflow runtime error in MSVC 2017. I do not want to check if the float a is small enough for a muliplication.

How is it possible to tell the C++ compiler to build a programm which ignores floating-point overflow and underflow errors? Researching for an answer I came across the function _controlfp_s at

https://learn.microsoft.com/de-de/cpp/c-runtime-library/reference/controlfp-s?view=vs-2019

But I did not manage to get it to work.

Warning C4756 is a compile time error and is letting you know that at compile time float b = a*a; will produce overflow. What value do you want to be stored into b (at compile time) in this case? — Richard Critten
– Richard Critten, Commented Nov 28, 2019 at 16:48
The above code snippet compiles just fine. The problem is that I am getting a runtime error. And I am asking how to set up the compiler so that the program will ignore the runtime error. The value to be stored in b is not relevant for me. — BlueTune
– BlueTune, Commented Nov 28, 2019 at 17:36
Warnings are not "compile time errors" @Richard. They can be made to act as errors but they are not errors. — Lightness Races in Orbit
– Lightness Races in Orbit, Commented Nov 28, 2019 at 17:53
The runtime error shown in the output reads: Exception thrown at 0x000000014012D7EC in Main.exe: 0xC0000091: Floating-point overflow (parameters: 0x0000000000000000, 0x00000000000011A8). Unhandled exception at 0x000000014012D7EC in Main.exe: 0xC0000091: Floating-point overflow (parameters: 0x0000000000000000, 0x00000000000011A8). — BlueTune
– BlueTune, Commented Nov 28, 2019 at 18:55

BlueTune · Accepted Answer · 2019-11-29 16:44:26Z

1

For Microsoft Visual C++ you can use _controlfp_s to get and to set the floating-point control word. For your code snippet a possible solution would look like:

int main()
{
    unsigned int fp_control;
    //Reading
    _controlfp_s(&fp_control, 0, 0);
    //Make changes
    unsigned int new_fp_control = fp_control | _EM_OVERFLOW | _EM_UNDERFLOW;
    //Update
    _controlfp_s(&fp_control, new_fp_control, _MCW_EM);

    float a = 68440675640679078541805800652800.0f;
    float b = a*a;
    out << b << ed;
}

answered Nov 29, 2019 at 16:44

BlueTune

1,0637 silver badges25 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Adrian Mole · Accepted Answer · 2019-11-28 16:56:30Z

0

You can disable the specific MSVC warning using the following #pragma (but it must be outside and before the main function):

#pragma warning(disable:4756)
void main()
{
   float a = 68440675640679078541805800652800.0f;
   float b = a*a;
   cout << b << endl;
}

The compiler will then silently give b the specific IEEE code that marks a floating-point value as "infinity" (see this Wikipedia page) and your cout line will reflect that.

At any point, you can re-enable the warning using a similar #pragma line but replacing "disable" with "default".

edited Nov 28, 2019 at 16:56

answered Nov 28, 2019 at 16:51

Adrian Mole

52.2k193 gold badges61 silver badges101 bronze badges

Collectives™ on Stack Overflow

Ignore floating-point overflow and underflow errors in C++

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related