/fp (Specify floating-point behavior)

Specifies how the compiler treats floating-point expressions, optimizations, and exceptions. The /fp options specify whether the generated code allows floating-point environment changes to the rounding mode, exception masks, and subnormal behavior, and whether floating-point status checks return current, accurate results. It controls whether the compiler generates code that maintains source operation and expression ordering and conforms to the standard for NaN propagation, or if it instead generates more efficient code that may reorder or combine operations and use simplifying algebraic transformations that are not allowed by the standard.

Syntax

/fp:[precise | strict | fast | except[-]]

Arguments

precise

By default, the compiler uses /fp:precise behavior.

Under /fp:precise the compiler preserves the source expression ordering and rounding properties of floating-point code when it generates and optimizes object code for the target machine. The compiler rounds to source code precision at four specific points during expression evaluation: at assignments, at typecasts, when a floating-point argument is passed to a function call, and when a floating-point value is returned from a function call. Intermediate computations may be performed at machine precision. Typecasts can be used to explicitly round intermediate computations.

The compiler does not perform algebraic transformations on floating-point expressions, such as reassociation or distribution, unless the transformation is guaranteed to produce a bitwise identical result. Expressions that involve special values (NaN, +infinity, -infinity, -0.0) are processed according to IEEE-754 specifications. For example, x != x evaluates to true if x is NaN. Floating-point contractions, that is, machine instructions that combine floating-point operations, may be generated under /fp:precise.

The compiler generates code intended to run in the default floating-point environment and assumes that the floating-point environment is not accessed or modified at runtime. That is, it assumes that the code does not unmask floating-point exceptions, read or write floating-point status registers, or change rounding modes.

If your floating-point code does not depend on the order of operations and expressions in your floating-point statements (for example, if you don't care whether a * b + a * c is computed as (b + c) * a or 2 * a as a + a), consider the /fp:fast option, which can produce faster, more efficient code. If your code both depends on the order of operations and expressions, and accesses or alters the floating-point environment (for example, to change rounding modes or to trap floating-point exceptions), use /fp:strict.

strict

/fp:strict has behavior similar to /fp:precise, that is, the compiler preserves the source ordering and rounding properties of floating-point code when it generates and optimizes object code for the target machine, and observes the standard when handling special values. In addition, the program may safely access or modify the floating-point environment at runtime.

Under /fp:strict, the compiler generates code that allows the program to safely unmask floating-point exceptions, read or write floating-point status registers, or change rounding modes. It rounds to source code precision at four specific points during expression evaluation: at assignments, at typecasts, when a floating-point argument is passed to a function call, and when a floating-point value is returned from a function call. Intermediate computations may be performed at machine precision. Typecasts can be used to explicitly round intermediate computations. The compiler does not perform algebraic transformations on floating-point expressions, such as reassociation or distribution, unless the transformation is guaranteed to produce a bitwise identical result. Expressions that involve special values (NaN, +infinity, -infinity, -0.0) are processed according to IEEE-754 specifications. For example, x != x evaluates to true if x is NaN. Floating-point contractions are not generated under /fp:strict.

/fp:strict is computationally more expensive than /fp:precise because the compiler must insert additional instructions to trap exceptions and allow programs to access or modify the floating-point environment at runtime. If your code doesn’t use this capability, but requires source code ordering and rounding, or relies on special values, use /fp:precise. Otherwise, consider using /fp:fast, which can produce faster and smaller code.

fast

The /fp:fast option allows the compiler to reorder, combine, or simplify floating-point operations to optimize floating-point code for speed and space. The compiler may omit rounding at assignment statements, typecasts, or function calls. It may reorder operations or perform algebraic transforms, for example, by use of associative and distributive laws, even if such transformations result in observably different rounding behavior. Because of this enhanced optimization, the result of some floating-point computations may differ from those produced by other /fp options. Special values (NaN, +infinity, -infinity, -0.0) may not be propagated or behave strictly according to the IEEE-754 standard. Floating-point contractions may be generated under /fp:fast. The compiler is still bound by the underlying architecture under /fp:fast, and additional optimizations may be available through use of the /arch option.

Under /fp:fast, the compiler generates code intended to run in the default floating-point environment and assumes that the floating-point environment isn’t accessed or modified at runtime. That is, it assumes that the code does not unmask floating-point exceptions, read or write floating-point status registers, or change rounding modes.

/fp:fast is intended for programs that do not require strict source code ordering and rounding of floating-point expressions, and do not rely on the standard rules for handling special values such as NaN. If your floating-point code requires preservation of source code ordering and rounding, or relies on standard behavior of special values, use /fp:precise. If your code accesses or modifies the floating-point environment to change rounding modes, unmask floating-point exceptions, or check floating-point status, use /fp:strict.

except

The /fp:except option generates code to ensures that any unmasked floating-point exceptions are raised at the exact point at which they occur, and that no additional floating-point exceptions are raised. By default, the /fp:strict option enables /fp:except, and /fp:precise does not. The /fp:except option is not compatible with /fp:fast. The option can be explicitly disabled by us of /fp:except-.

Note that /fp:except does not enable any floating-point exceptions by itself, but it is required for programs to enable floating-point exceptions. See _controlfp for information on how to enable floating-point exceptions.

Remarks

Multiple /fp options can be specified in the same compiler command line. Only one of /fp:strict, /fp:fast, and /fp:precise options can be in effect at a time. If more than one of these options is specified on the command line, the later option takes precedence and the compiler generates a warning. The /fp:strict and /fp:except options are not compatible with /clr.

The /Za (ANSI compatibility) option is not compatible with /fp.

Using Pragmas to Control Floating-Point Behavior

The compiler provides three pragma directivess to override the floating-point behavior specified on the command-line: float_control, fenv_access, and fp_contract. You can use these pragmas to control floating-point behavior at function-level, not within a function. Note that these pragmas do not correspond directly to the /fp options. This table shows how the /fp options and pragmas map to each other. For more information, see the documentation for the individual options and pragmas.

float_control(precise) float_control(except) fenv_access fp_contract
/fp:fast off off off on
/fp:precise on off off on
/fp:except on on on off

The default floating point environment

When a process is initialized, the default floating point environment is set. This environment masks all floating point exceptions, sets the rounding mode to round to nearest (FE_TONEAREST), preserves subnormal (denormal) values, uses the default precision of significand (mantissa) for float, double, and long double values, and where supported, sets the infinity control to the default affine mode.

Floating-point environment access and modification

The Microsoft Visual C++ runtime provides several functions to access and modify the floating-point environment. These include _controlfp, _clearfp, and _statusfp and their variants. To ensure correct program behavior when your code accesses or modifies the floating-point environment, fenv_access must be enabled, either by the /fp:strict option or by use of the fenv_access pragma, for these functions to have any effect. When fenv_access is not enabled, access or modification of the floating-point environment may result in unexpected program behavior: code may not honor requested changes to the floating-point environment; the floating-point status registers may not report expected or current results; and unexpected floating-point exceptions may occur or expected floating-point exceptions may not occur.

When your code accesses or modifies the floating-point environment, you must be careful when you combine code where fenv_access is enabled with code that does not have fenv_access enabled. In code where fenv_access is not enabled, the compiler assumes that the platform default floating-point environment is in effect, and that the floating-point status is not accessed or modified. We recommend you save and restore the local floating-point environment to its default state before control is transferred to a function that does not have fenv_access enabled. This example demonstrates how the float_control pragma can be set and restored:

#pragma float_control(strict, on, push)
// Code that uses /fp:strict mode
#pragma float_control(pop)

Floating-point rounding modes

Under both /fp:precise and /fp:fast the compiler generates code intended to run in the default floating-point environment and assumes that the environment isn’t accessed or modified at runtime. That is, it assumes that the code does not unmask floating-point exceptions, read or write floating-point status registers, or change rounding modes. However, some programs need to alter the floating-point environment. For example, this sample computes error bounds of a floating-point multiplication by altering floating-point rounding modes:

// fp_error_bounds.cpp
#include <iostream>
#include <limits>
using namespace std;

int main(void)
{
    float a = std::<float>::max();
    float b = -1.1;
    float cLower = 0.0;
    float cUpper = 0.0;
    unsigned int control_word = 0;
    int err = 0;

    // compute lower error bound.
    // set rounding mode to -infinity.
    err = _controlfp_s(&control_word, _RC_DOWN, _MCW_RC);
    if (err)
    {
        cout << "_controlfp_s(&control_word, _RC_DOWN, _MCW_RC) failed with error:" << err << endl;
    }  
    cLower = a * b;

    // compute upper error bound.
    // set rounding mode to +infinity.
    err = _controlfp_s(&control_word, _RC_UP, _MCW_RC);
    if (err)
    {
        cout << "_controlfp_s(&control_word, _RC_UP, _MCW_RC) failed with error:" << err << endl;
    }
    cUpper = a * b;

    // restore default rounding mode.
    err = _controlfp_s(&control_word, _CW_DEFAULT, _MCW_RC);
    if (err)
    {
        cout << "_controlfp_s(&control_word, _CW_DEFAULT, _MCW_RC) failed with error:" << err << endl;
    }
    // display error bounds.
    cout << "cLower = " << cLower << endl;
    cout << "cUpper = " << cUpper << endl;
    return 0;
}

Since the compiler assumes the default floating point environment under /fp:fast and /fp:precise it is free to ignore the calls to _controlfp_s. For example, when compiled by using both /O2 and /fp:precise for the x86 architecture, the bounds are not computed, and the sample program outputs:

cLower = -inf
cUpper = -inf

When compiled with both /O2 and /fp:strict for the x86 architecture, the sample program outputs:

cLower = -inf
cUpper = -3.40282e+38

Floating-point special values

Under /fp:precise and /fp:strict, expressions that involve special values (NaN, +infinity, -infinity, -0.0) behave according to the IEEE-754 specifications. Under /fp:fast, the behavior of these special values may be inconsistent with IEEE-754.

This sample demonstrates the different behavior of special values under /fp:precise, /fp:strict and /fp:fast:

// fp_special_values.cpp
#include <stdio.h>
#include <cmath>

float gf0 = -0.0;

int main()
{
    float f1 = INFINITY;
    float f2 = NAN;
    float f3 = -INFINITY;
    bool a, b;
    float c, d, e;
    a = (f1 == f1);
    b = (f2 == f2);
    c = (f1 - f1);
    d = (f2 - f2);
    printf("INFINITY == INFINITY : %d\n", a);
    printf("NAN == NAN           : %d\n", b);
    printf("INFINITY - INFINITY  : %f\n", c);
    printf("NAN - NAN            : %f\n", d);

    e = gf0 / abs(f3);
    printf("std::signbit(-0.0/-INFINITY): %d\n", std::signbit(c));
    return 0;
}

When compiled with /O2 /fp:precise or /O2 /fp:strict for x86 architecture, the outputs are consistent with the IEEE-754 specification:

INFINITY == INFINITY : 1
NAN == NAN           : 0
INFINITY - INFINITY  : -nan(ind)
NAN - NAN            : -nan(ind)
std::signbit(-0.0/-INFINITY): 1

When compiled with /O2 /fp:fast for x86 architecture, the outputs are not consistent with IEEE-754:

INFINITY == INFINITY : 1
NAN == NAN           : 1
INFINITY - INFINITY  : 0.000000
NAN - NAN            : 0.000000
std::signbit(-0.0/-INFINITY): 0

Floating-point algebraic transformations

Under /fp:precise and /fp:strict, the compiler does not perform mathematical transformations unless the transformation is guaranteed to produce a bitwise identical result. The compiler may perform such transformations under /fp:fast. For example, the expression a * b + a * c in the sample function algebraic_transformation may be compiled into a * (b + c) under /fp:fast. Such transformations aren’t performed under /fp:precise or /fp:strict, and the compiler generates a * b + a * c.

float algebraic_transformation (float a, float b, float c)
{
    return a * b + a * c;
}

Floating-point explicit casting points

Under /fp:precise and /fp:strict, the compiler rounds to source code precision at four specific points during expression evaluation: at assignments, at typecasts, when a floating-point argument is passed to a function call, and when a floating-point value is returned from a function call. Typecasts can be used to explicitly round intermediate computations. Under /fp:fast, the compiler does not generate explicit casts at these points to guarantee source code precision. This sample demonstrates the behavior under different /fp options:

float casting(float a, float b)
{
    return 5.0*((double)(a+b));
}

When compiled by using /O2 /fp:precise or /O2 /fp:strict, you can see that explicit type casts are inserted at both the typecast and at the function return point in the generated code for the x64 architecture:

        addss    xmm0, xmm1
        cvtss2sd xmm0, xmm0
        mulsd    xmm0, QWORD PTR __real@4014000000000000
        cvtsd2ss xmm0, xmm0
        ret      0

Under /O2 /fp:fast the generated code is simplified, because all type casts are optimized away:

        addss    xmm0, xmm1
        mulss    xmm0, DWORD PTR __real@40a00000
        ret      0

To set this compiler option in the Visual Studio development environment

  1. Open the project's Property Pages dialog box. For details, see Working with Project Properties.

  2. Select the Configuration Properties > C/C++ > Code Generation property page.

  3. Modify the Floating Point Model property.

To set this compiler option programmatically

See also

Compiler Options
Setting Compiler Options
Microsoft Visual C++ Floating Point Optimization