Report a bug
If you spot a problem with this page, click here to create a Bugzilla issue.
Improve this page
Quickly fork, edit online, and submit a pull request for this page. Requires a signed-in GitHub account. This works well for small changes. If you'd like to make larger changes you may want to consider using a local clone.

std.math.hardware

This is a submodule of std.math.
It contains hardware support for floating point numbers.
Authors:
Walter Bright, Don Clugston, Conversion of CEPHES math library to D by Iain Buclaw and David Nadlinger
struct IeeeFlags;
IEEE exception status flags ('sticky bits')
These flags indicate that an exceptional floating-point condition has occurred. They indicate that a NaN or an infinity has been generated, that a result is inexact, or that a signalling NaN has been encountered. If floating-point exceptions are enabled (unmasked), a hardware exception will be generated instead of setting these flags.
Examples:
import std.math.traits : isNaN;

static void func() {
    int a = 10 * 10;
}
pragma(inline, false) static void blockopt(ref real x) {}
real a = 3.5;
// Set all the flags to zero
resetIeeeFlags();
assert(!ieeeFlags.divByZero);
blockopt(a); // avoid constant propagation by the optimizer
// Perform a division by zero.
a /= 0.0L;
writeln(a); // real.infinity
assert(ieeeFlags.divByZero);
blockopt(a); // avoid constant propagation by the optimizer
// Create a NaN
a *= 0.0L;
assert(ieeeFlags.invalid);
assert(isNaN(a));

// Check that calling func() has no effect on the
// status flags.
IeeeFlags f = ieeeFlags;
func();
writeln(ieeeFlags); // f
const nothrow @nogc @property @safe bool inexact();
The result cannot be represented exactly, so rounding occurred.

Example x = sin(0.1);

const nothrow @nogc @property @safe bool underflow();
A zero was generated by underflow

Example x = real.min*real.epsilon/2;

const nothrow @nogc @property @safe bool overflow();
An infinity was generated by overflow

Example x = real.max*2;

const nothrow @nogc @property @safe bool divByZero();
An infinity was generated by division by zero

Example x = 3/0.0;

const nothrow @nogc @property @safe bool invalid();
A machine NaN was generated.

Example x = real.infinity * 0.0;

nothrow @nogc @trusted void resetIeeeFlags();
Set all of the floating-point status flags to false.
Examples:
pragma(inline, false) static void blockopt(ref real x) {}
resetIeeeFlags();
real a = 3.5;
blockopt(a); // avoid constant propagation by the optimizer
a /= 0.0L;
blockopt(a); // avoid constant propagation by the optimizer
writeln(a); // real.infinity
assert(ieeeFlags.divByZero);

resetIeeeFlags();
assert(!ieeeFlags.divByZero);
pure nothrow @nogc @property @trusted IeeeFlags ieeeFlags();
Returns:
snapshot of the current state of the floating-point status flags
Examples:
import std.math.traits : isNaN;

pragma(inline, false) static void blockopt(ref real x) {}
resetIeeeFlags();
real a = 3.5;
blockopt(a); // avoid constant propagation by the optimizer

a /= 0.0L;
writeln(a); // real.infinity
assert(ieeeFlags.divByZero);
blockopt(a); // avoid constant propagation by the optimizer

a *= 0.0L;
assert(isNaN(a));
assert(ieeeFlags.invalid);
struct FloatingPointControl;
Control the Floating point hardware
Change the IEEE754 floating-point rounding mode and the floating-point hardware exceptions.
By default, the rounding mode is roundToNearest and all hardware exceptions are disabled. For most applications, debugging is easier if the division by zero, overflow, and invalid operation exceptions are enabled. These three are combined into a severeExceptions value for convenience. Note in particular that if invalidException is enabled, a hardware trap will be generated whenever an uninitialized floating-point variable is used.
All changes are temporary. The previous state is restored at the end of the scope.

Example

{
    FloatingPointControl fpctrl;

    // Enable hardware exceptions for division by zero, overflow to infinity,
    // invalid operations, and uninitialized floating-point variables.
    fpctrl.enableExceptions(FloatingPointControl.severeExceptions);

    // This will generate a hardware exception, if x is a
    // default-initialized floating point variable:
    real x; // Add `= 0` or even `= real.nan` to not throw the exception.
    real y = x * 3.0;

    // The exception is only thrown for default-uninitialized NaN-s.
    // NaN-s with other payload are valid:
    real z = y * real.nan; // ok

    // The set hardware exceptions and rounding modes will be disabled when
    // leaving this scope.
}

Examples:
import std.math.rounding : lrint;

FloatingPointControl fpctrl;

fpctrl.rounding = FloatingPointControl.roundDown;
writeln(lrint(1.5)); // 1.0

fpctrl.rounding = FloatingPointControl.roundUp;
writeln(lrint(1.4)); // 2.0

fpctrl.rounding = FloatingPointControl.roundToNearest;
writeln(lrint(1.5)); // 2.0
alias RoundingMode = uint;
roundToNearest

roundDown

roundUp

roundToZero

roundingMask
IEEE rounding modes. The default mode is roundToNearest.
roundingMask = A mask of all rounding modes.
nothrow @nogc @property @trusted void rounding(RoundingMode newMode);
Change the floating-point hardware rounding mode
Changing the rounding mode in the middle of a function can interfere with optimizations of floating point expressions, as the optimizer assumes that the rounding mode does not change. It is best to change the rounding mode only at the beginning of the function, and keep it until the function returns. It is also best to add the line:
pragma(inline, false);
as the first line of the function so it will not get inlined.
Parameters:
RoundingMode newMode the new rounding mode
static pure nothrow @nogc @property @trusted RoundingMode rounding();
Returns:
the currently active rounding mode
alias ExceptionMask = uint;
subnormalException

inexactException

underflowException

overflowException

divByZeroException

invalidException

severeExceptions

allExceptions
IEEE hardware exceptions. By default, all exceptions are masked (disabled).
severeExceptions = The overflow, division by zero, and invalid exceptions.
static pure nothrow @nogc @property @safe bool hasExceptionTraps();
Returns:
true if the current FPU supports exception trapping
nothrow @nogc @trusted void enableExceptions(ExceptionMask exceptions);
Enable (unmask) specific hardware exceptions. Multiple exceptions may be ORed together.
nothrow @nogc @trusted void disableExceptions(ExceptionMask exceptions);
Disable (mask) specific hardware exceptions. Multiple exceptions may be ORed together.
static pure nothrow @nogc @property @trusted ExceptionMask enabledExceptions();
Returns:
the exceptions which are currently enabled (unmasked)