WMMX General Support Intrinsics (Windows Embedded CE 6.0)
1/5/2010
The intrinsic functions listed in the following table provide general support operations.
Intrinsic name |
Operation |
WMMX Instruction |
_mm_extract_pi8 |
Extracts on 8 bytes |
TEXTRMSB |
_mm_extract_pi16 |
Extracts on 4 half words |
TEXTRMSH |
_mm_extract_pi32 |
Extracts on two words |
TEXTRMSW |
_mm_extract_pu8 |
Extracts on 8 bytes |
TEXTRMUB |
_mm_extract_pu16 |
Extracts on 4 half words |
TEXTRMUL |
_mm_extract_pu32 |
Extracts on two words |
TEXTRMUW |
_mm_insert_pi8 |
Inserts a byte. |
TINSRB |
_mm_insert_pi16 |
Inserts a half word |
TINSRH |
_mm_insert_pi32 |
Inserts a word |
TINSRW |
_mm_max_pi8 |
Computes maximum |
WMAXSB |
_mm_max_pi16 |
Computes maximum |
WMAXSH |
_mm_max_pi32 |
Computes maximum |
WMAXUB |
_mm_max_pu8 |
Computes maximum, unsigned |
WMAXUB |
_mm_max_pu16 |
Computes maximum, unsigned |
WMAXUH |
_mm_max_pu32 |
Computes maximum, unsigned |
WMAXUW |
_mm_min_pi8 |
Computes minimum |
WMINXSB |
_mm_min_pi16 |
Computes minimum |
WMINSH |
_mm_min_pi32 |
Computes minimum |
WMINUB |
_mm_min_pu8 |
Computes minimum, unsigned |
WMINUB |
_mm_min_pu16 |
Computes minimum, unsigned |
WMINUH |
_mm_min_pu32 |
Computes minimum, unsigned |
WMINUW |
_mm_movemask_pi8 |
Creates an 8-bit mask |
TMOVMSKB |
_mm_movemask_pi16 |
Creates a 4 half-word mask |
TMOVMSKH |
_mm_movemask_pi32 |
Creates a two word mask |
TMOVMSKW |
_mm_shuffle_pi8 |
Returns a combination of 4 words |
WSHUFH |
_mm_avg_pu8 |
Computes rounded average |
WAV2BR |
_mmavg_pu16 |
Computes rounded average |
WAV2HR |
_mmavg2_pu8 |
Computes average |
WAVG2B |
_mmavg2_pu16 |
Computes average |
WAVG2H |
_mm_sada_pu8 |
Computes absolute sum of differences |
WSADB |
_mm_sada_pu16 |
Computes absolute sum of differences |
WSADH |
_mm_align_si64 |
Extracts a 64-bit value |
WALIGN/WALIGNR |
_mm_cvtsi32_si64 |
Converts from int to a 64-bit __m64 object |
TMCRR |
_mm_cvtsi64_si32 |
Converts from__m64 object to int |
TMRRC |
_mm_empty |
Empties MM state |
--- |