MMX(TM) Technology Intrinsics Implementation

Key to the table entries

A = Expected to give significant performance gain over non-intrinsic-based code equivalent.
B = Non-intrinsic-based source code would be better; the intrinsic's implementation may map directly to native instructions, but they offer no significant performance gain.
C = Requires contorted implementation for particular microarchitecture. Will result in very poor performance if used.

Intrinsic Name	Alternate Name	Across All IA	MMX(TM) Technology Streaming SIMD Extensions Streaming SIMD Extensions 2	Itanium® Architecture
_m_empty	_mm_empty	N/A	A	B
_m_from_int	_mm_cvtsi32_si64	N/A	A	A
_m_to_int	_mm_cvtsi64_si32	N/A	A	A
_m_packsswb	_mm_packs_pi16	N/A	A	A
_m_packssdw	_mm_packs_pi32	N/A	A	A
_m_packuswb	_mm_packs_pu16	N/A	A	A
_m_punpckhbw	_mm_unpackhi_pi8	N/A	A	A
_m_punpckhwd	_mm_unpackhi_pi16	N/A	A	A
_m_punpckhdq	_mm_unpackhi_pi32	N/A	A	A
_m_punpcklbw	_mm_unpacklo_pi8	N/A	A	A
_m_punpcklwd	_mm_unpacklo_pi16	N/A	A	A
_m_punpckldq	_mm_unpacklo_pi32	N/A	A	A
_m_paddb	_mm_add_pi8	N/A	A	A
_m_paddw	_mm_add_pi16	N/A	A	A
_m_paddd	_mm_add_pi32	N/A	A	A
_m_paddsb	_mm_adds_pi8	N/A	A	A
_m_paddsw	_mm_adds_pi16	N/A	A	A
_m_paddusb	_mm_adds_pu8	N/A	A	A
_m_paddusw	_mm_adds_pu16	N/A	A	A
_m_psubb	_mm_sub_pi8	N/A	A	A
_m_psubw	_mm_sub_pi16	N/A	A	A
_m_psubd	_mm_sub_pi32	N/A	A	A
_m_psubsb	_mm_subs_pi8	N/A	A	A
_m_psubsw	_mm_subs_pi16	N/A	A	A
_m_psubusb	_mm_subs_pu8	N/A	A	A
_m_psubusw	_mm_subs_pu16	N/A	A	A
_m_pmaddwd	_mm_madd_pi16	N/A	A	C
_m_pmulhw	_mm_mulhi_pi16	N/A	A	A
_m_pmullw	_mm_mullo_pi16	N/A	A	A
_m_psllw	_mm_sll_pi16	N/A	A	A
_m_psllwi	_mm_slli_pi16	N/A	A	A
_m_pslld	_mm_sll_pi32	N/A	A	A
_m_pslldi	_mm_slli_pi32	N/A	A	A
_m_psllq	_mm_sll_si64	N/A	A	A
_m_psllqi	_mm_slli_si64	N/A	A	A
_m_psraw	_mm_sra_pi16	N/A	A	A
_m_psrawi	_mm_srai_pi16	N/A	A	A
_m_psrad	_mm_sra_pi32	N/A	A	A
_m_psradi	_mm_srai_pi32	N/A	A	A
_m_psrlw	_mm_srl_pi16	N/A	A	A
_m_psrlwi	_mm_srli_pi16	N/A	A	A
_m_psrld	_mm_srl_pi32	N/A	A	A
_m_psrldi	_mm_srli_pi32	N/A	A	A
_m_psrlq	_mm_srl_si64	N/A	A	A
_m_psrlqi	_mm_srli_si64	N/A	A	A
_m_pand	_mm_and_si64	N/A	A	A
_m_pandn	_mm_andnot_si64	N/A	A	A
_m_por	_mm_or_si64	N/A	A	A
_m_pxor	_mm_xor_si64	N/A	A	A
_m_pcmpeqb	_mm_cmpeq_pi8	N/A	A	A
_m_pcmpeqw	_mm_cmpeq_pi16	N/A	A	A
_m_pcmpeqd	_mm_cmpeq_pi32	N/A	A	A
_m_pcmpgtb	_mm_cmpgt_pi8	N/A	A	A
_m_pcmpgtw	_mm_cmpgt_pi16	N/A	A	A
_m_pcmpgtd	_mm_cmpgt_pi32	N/A	A	A
_mm_setzero_si64		N/A	A	A
_mm_set_pi32		N/A	A	A
_mm_set_pi16		N/A	A	C
_mm_set_pi8		N/A	A	C
_mm_set1_pi32		N/A	A	A
_mm_set1_pi16		N/A	A	A
_mm_set1_pi8		N/A	A	A
_mm_setr_pi32		N/A	A	A
_mm_setr_pi16		N/A	A	C
_mm_setr_pi8		N/A	A	C

_mm_empty is implemented in Itanium instructions as a NOP for source compatibility only.