3DNow! Intrinsics

2011-07-25

Microsoft Specific

This topic contains the 3DNow! intrinsics. For each intrinsic, the header file mm3dnow.h is required.

The following table lists the 3DNow! intrinsics alphabetically.

Intrinsic	Use
_m_femms	Clears the architectural state when switching between MMX and floating-point instructions.
_m_from_float	Returns a 64-bit MMX value where the lower half is set to the floating-point, single-precision value from the source operand and the upper half is zero. There is no error return.
_m_pavgusb	Calculates the rounded averages of eight unsigned 8-bit integer values.
_m_pf2id	Converts packed floating-point, single-precision values to packed 32-bit integer values.
_m_pf2iw	Converts packed floating-point, single-precision values to packed 16-bit signed integer values using truncation.
_m_pfacc	Performs packed floating-point, single-precision accumulation.
_m_pfadd	Performs packed floating-point, single-precision addition.
_m_pfcmpeq	Compares packed floating-point, single-precision values to be equal and sets the corresponding return value to ones or zeros based on the result of the comparison.
_m_pfcmpge	Compares the first packed floating-point, single-precision value to be greater than or equal to the second one and sets the corresponding return value to ones or zeros based on the result of the comparison.
_m_pfcmpgt	Compares the first packed floating-point, single-precision value to be greater than the second one and sets the corresponding return value to ones or zeros based on the result of the comparison.
_m_pfmax	Returns the larger of the two packed floating-point, single-precision values.
_m_pfmin	Returns the smaller of the two packed floating-point, single-precision values.
_m_pfmul	Performs packed floating-point, single-precision multiplication.
_m_pfnacc	Performs packed floating-point, single-precision negative accumulation.
_m_pfpnacc	Performs packed floating-point, single-precision positive-negative accumulation.
_m_pfrcp	Performs scalar floating-point, low-precision reciprocal approximation.
_m_pfrcpit1	Performs the first intermediate step in the Newton-Raphson iteration to refine the reciprocal approximation produced by the _m_pfrcp intrinsic function.
_m_pfrcpit2	Performs the second and final step in the Newton-Raphson iteration to refine the reciprocal or reciprocal square root approximation produced by the _m_pfrcp or _m_pfsqrt intrinsic functions, respectively.
_m_pfrsqit1	Performs the first intermediate step in the Newton-Raphson iteration to refine the reciprocal square root approximation produced by _m_pfsqrt intrinsic function.
_m_pfrsqrt	Performs scalar floating-point, low-precision reciprocal square root approximation.
_m_pfsub	Performs packed floating-point, single-precision subtraction.
_m_pfsubr	Performs packed floating-point, single-precision reverse subtraction.
_m_pi2fd	Converts packed 32-bit integer values to packed floating-point, single-precision values.
_m_pi2fw	Converts packed 16-bit signed integer values to packed floating-point, single-precision values.
_m_pmulhrw	Multiplies four signed 16-bit integer values in the source operand by four signed 16-bit integer values in the destination operand.
_m_prefetch	Loads a 32-byte cache line into L1 data cache and sets the cache line state to exclusive.
_m_prefetchw	Loads a 32-byte cache line into L1 data cache and sets the cache line state to modified.
_m_pswapd	Swaps upper and lower halves of the source operand.
_m_to_float	Returns the floating-point, single-precision value from the lower half of the 64-bit MMX value in the source operand. There is no error return. The compiler correctly ensures that an implict FEMMS is issued before any attempt to use the result of _m_to_float( ) operation.

通过

3DNow! Intrinsics

See Also

Reference

其他资源