fp-trap-all, Qfp-trap-all

Architecture Restrictions

Not available on Intel® 64 architecture targeting the Intel® Xeon Phi™ coprocessor x100 product family (formerly code name Knights Corner), on IA-32 architecture targeting Intel® Graphics Technology, or on Intel® 64 architecture targeting Intel® Graphics Technology

Syntax

Linux and OS X:

-fp-trap-all=mode[,mode,...]

Windows:

/Qfp-trap-all:mode[,mode,...]

Arguments

mode

Is the floating-point trapping mode. If you specify more than one mode value, the list is processed sequentially from left to right. Possible values are:

[no]divzero	Enables or disables the IEEE trap for division by zero.
[no]inexact	Enables or disables the IEEE trap for inexact result.
[no]invalid	Enables or disables the IEEE trap for invalid operation.
[no]overflow	Enables or disables the IEEE trap for overflow.
[no]underflow	Enables or disables the IEEE trap for underflow.
[no]denormal	Enables or disables the trap for denormal.
all	Enables all of the above traps.
none	Disables all of the above traps.
common	Sets the most commonly used IEEE traps: division by zero, invalid operation, and overflow.

Default

-fp-trap-all=none
or/Qfp-trap-all:none

No traps are enabled for all routines.

Description

This option sets the floating-point trapping mode for the main routine. It does not set a handler for floating-point exceptions.

The [no] form of a mode value is only used to modify the meaning of mode values all and common, and can only be used with one of those values. The [no] form of the option by itself does not explicitly cause a particular trap to be disabled.

Use mode value inexact with caution. This results in the trap being enabled whenever a floating-point value cannot be represented exactly, which can cause unexpected results.

If mode value underflow is specified, the compiler ignores the FTZ (flush-to-zero) bit state of Intel® Streaming SIMD Extensions (Intel® SSE) floating-point units.

When a DAZ (denormals are zero) bit is set in an Intel® SSE floating-point unit control word, a denormal operand exception is never generated.

To set the floating-point trapping mode for the main routine only, specify the [Q]fp-trap option.

Note

The negative form of the [Q]ftz option can be used to set or reset the FTZ and the DAZ hardware flags.

IDE Equivalent

None

Alternate Options