HP MLIB User's Guide Vol. 2 7th Ed.

Chapter 12 VMATH 803

Set and get the mode for VMATH functions vmlSetMode, vmlGetMode

Name vmlSetMode, vmlGetMode

Set and get the mode for VMATH functions

Purpose vmlSetMode sets the new mode for VMATH functions according to the mode

parameter and returns the mode in effect at the time of the call.

vmlGetMode returns the current VMATH mode (as does vmlSetMode).

The mode for the VMATH library comprises an accuracy mode, an error mode,

and an FPU mode.

The accuracy mode is speciﬁed by one of the macros:

The error mode is speciﬁed by one of the macros:

The FPU mode has no effect in VMATH. However, for user source compatibility

with Intel’s MKL vector math library, an FPU mode can be speciﬁed by one of

the macros:

VML_FLOAT_CONSISTENT

VML_DOUBLE_CONSISTENT

VML_RESTORE

The mode parameter for vmlSetMode is interpreted as a bitwise of OR one or

more of an accuracy mode macro, and error mode macro, and an FPU mode

macro. If an accuracy mode, error mode, or FPU mode is not represented in the

mode parameter, then that mode is unchanged by the call.

For example, the following saves the current mode, installs the modes for lower

accuracy and ignoring errors, while leaving the FPU mode unchanged, and

then restores the saved mode:

savemode = vmlSetMode(VML_LA | VML_ERRMODE_IGNORE);

VML_HA Higher accuracy versions of VMATH functions

are used (default)

VML_LA Lower accuracy versions of VMATH functions

may be used to increase throughput

VML_ERRMODE_EXCEPT Floating-point exceptions, in the style of the

IEEE 754 and C99 standards, are signaled

VML_ERRMODE_IGNORE No error reporting is in effect

VML_ERRMODE_DEFAULT Floating-point exceptions are raised