/fpe

Syntax:

/fpe:level

The /fpe:level option controls floating-point exception handling at run time for the main program. This includes whether exceptional floating-point values are allowed and how precisely run-time exceptions are reported. The /fpe:level option specifies how the compiler should handle the following floating-point exceptions:

When floating-point calculations result in a divide by zero, overflow, or invalid operation.
When floating-point calculations result in an underflow.
When a denormalized number or other exceptional number (positive infinity, negative infinity, or a NaN) is present in an arithmetic expression

For performance reasons:

On ia32 systems, the default is /fpe:3. Using /fpe:0 will slow run-time performance on ia32 systems.
On ia64 systems, the default is /fpe:0 (many programs do not need to handle denormalized numbers or other exceptional values). Using /fpe:3 will slow run-time performance on ia64 systems.

In the visual development environment, specify the Floating-Point Exception Handling in the Floating Point Compiler Option Category. The /fpe:level (level is 0 or 3) options are as follows:

/fpe:level Option	Handling of Underflow	Handling of Divide by Zero, Overflow, and Invalid Operation
/fpe:0 (default on ia64 systems)	Sets any calculated denormalized value (result) to zero and lets the program continue. Any use of a denormalized number in an arithmetic expression uses a value of 0.0 and execution continues. A message is displayed only if /check:underflow is also specified.	Exceptional values are not allowed. The program terminates after displaying a message.
/fpe:3 (default on ia32 systems)	Leaves any calculated denormalized value as is. The program continues, allowing gradual underflow. Use of a denormalized (or exceptional) number in an arithmetic expression results in program continuation, but with slower performance.	The program continues. No message is displayed. A NaN or Infinity (+ or -) will be generated.

When compiling different routines in a program separately, you should use the same /fpe:level value.

On ia32 systems, the /fpe option, /check:underflow option, and MATHERRQQ routine interact as follows:

Specified /fpe:n Option	Was /check:underflow Specified?	Is a User-Written MATHERRQQ Routine Present?	Underflow Handling by the Visual Fortran Run-Time System on ia32 Systems
/fpe:0	No	No	The underflowed result is set to zero (0). The program continues.
/fpe:0	No	Yes	The underflowed result is set to zero (0). The program continues.
/fpe:0	Yes	No	The underflowed result is set to zero (0). The program continues. The number of underflowed results are counted and messages are displayed for the first two occurrences.
/fpe:0	Yes	Yes	The underflowed result is set to zero (0). The program continues. The number of underflowed results are counted and messages are displayed for the first two occurrences.
/fpe:3	No	No	Denormalized results are allowed and the program continues. Traps are masked and no handlers are invoked.
/fpe:3	No	Yes	Denormalized results are allowed and the program continues. Traps are masked and no handlers are invoked.
/fpe:3	Yes	No	For Version 6, a fatal error results and the program terminates.
/fpe:3	Yes	Yes	Depends on the source causing the underflow: If the underflow occurs in an intrinsic procedure, the undefined result is left as is. The program continues with the assumption that the user-specified MATHERRQQ handler will perform any result fix up needed. If the underflow does not occur in an intrinsic procedure, for Version 6.0, a fatal error results and the program terminates.

For more information:

About the /fpe option on ia32 systems, see How the Floating-Point Exception Handling (/fpe) Compiler Option Works.
About the floating-point environment and the MATHERRQQ routine (ia32 systems), see The Floating-Point Environment.
About creating exception and termination handlers, see Advanced Exception and Termination Handling Considerations.
About routines that can obtain or set the floating-point exception settings used by Visual Fortran at run-time, see FOR_SET_FPE and FOR_GET_FPE.
About IEEE floating-point exception handling, see the IEEE Standard for Binary Floating-Point Arithmetic (ANSI/IEEE Standard 754-1985).