Boost C++ Libraries Home Libraries People FAQ More

PrevUpHomeNext

Configuration Macros

Almost all configuration details are set up automatically by <boost\math\tools\config.hpp>.

In normal use, only policy configuration macros are likely to be used. See policy reference.

For reference, information on Boost.Math macros used internally are described briefly below.

Table 1.11. Boost.Math Macros

MACRO

Notes

BOOST_MATH_NO_LONG_DOUBLE_MATH_FUNCTIONS

Do not produce or use long double functions: this macro gets set when the platform's long double or standard library long double support is absent or buggy.

BOOST_MATH_USE_FLOAT128

When set the numeric constants support the __float128 data type with constants having the Q suffix.

BOOST_MATH_DISABLE_FLOAT128

When set the numeric constants do not use the __float128 data type even if the compiler appears to support it.

BOOST_MATH_NO_REAL_CONCEPT_TESTS

Do not try to use real concept tests (hardware or software does not support real_concept type).

BOOST_MATH_CONTROL_FP

Controls FP hardware exceptions - our tests don't support hardware exceptions on MSVC. May get set to something like: _control87(MCW_EM,MCW_EM).

BOOST_MATH_NO_DEDUCED_FUNCTION_POINTERS

This macro is used by our test cases, it is set when an assignment of a function template to a function pointer requires explicit template arguments to be provided on the function name.

BOOST_MATH_USE_C99

Use C99 math functions.

BOOST_NO_NATIVE_LONG_DOUBLE_FP_CLASSIFY

define if no native (or buggy) fpclassify(long double) even though the other C99 functions are present.

BOOST_MATH_SMALL_CONSTANT(x)

Helper macro used in our test cases to set underflowing constants set to zero if this would cause compiler issues.

BOOST_MATH_BUGGY_LARGE_FLOAT_CONSTANTS

Set if constants too large for a float, will cause "bad" values to be stored in the data, rather than infinity or a suitably large value.

BOOST_MATH_STD_USING

Provides using statements for many std:: (abs to sqrt) and boost::math (rounds, modf) functions. This allows these functions to be called unqualified so that if argument-dependent Argument Dependent Lookup fails to find a suitable overload, then the std:: versions will also be considered.

BOOST_FPU_EXCEPTION_GUARD

Used at the entrypoint to each special function to reset all FPU exception flags prior to internal calculations, and then merge the old and new exception flags on function exit. Used as a workaround on platforms or hardware that behave strangely if any FPU exception flags are set when calling standard library functions.

BOOST_MATH_INSTRUMENT

Define to output diagnostics for math functions. This is rather 'global' to Boost.Math and so coarse-grained that it will probably produce copious output! (Especially because full precision values are output). Designed primarily for internal use and development.

BOOST_MATH_INSTRUMENT_CODE(x)

Output selected named variable, for example BOOST_MATH_INSTRUMENT_CODE("guess = " << guess); Used by BOOST_MATH_INSTRUMENT

BOOST_MATH_INSTRUMENT_VARIABLE(name)

Output selected variable, for example BOOST_MATH_INSTRUMENT_VARIABLE(result); Used by BOOST_MATH_INSTRUMENT

BOOST_MATH_INSTRUMENT_FPU

Output the state of the FPU's control flags.


Table 1.12. Boost.Math Tuning

Macros for Tuning performance options for specific compilers

Notes

BOOST_MATH_POLY_METHOD

See the performance tuning section.

BOOST_MATH_RATIONAL_METHOD

See the performance tuning section.

BOOST_MATH_MAX_POLY_ORDER

See the performance tuning section.

BOOST_MATH_INT_TABLE_TYPE

See the performance tuning section.

BOOST_MATH_INT_VALUE_SUFFIX

Helper macro for appending the correct suffix to integer constants which may actually be stored as reals depending on the value of BOOST_MATH_INT_TABLE_TYPE.



PrevUpHomeNext