Apply the right patch.

From-SVN: r67544
This commit is contained in:
H.J. Lu 2003-06-06 07:06:41 -07:00
parent 30fb323110
commit c220e3a948
2 changed files with 447 additions and 81 deletions

View File

@ -48,6 +48,12 @@ _mm_empty (void)
__builtin_ia32_emms ();
}
static __inline void
_m_empty (void)
{
_mm_empty ();
}
/* Convert I to a __m64 object. The integer is zero-extended to 64-bits. */
static __inline __m64
_mm_cvtsi32_si64 (int __i)
@ -56,6 +62,12 @@ _mm_cvtsi32_si64 (int __i)
return (__m64) __tmp;
}
static __inline __m64
_m_from_int (int __i)
{
return _mm_cvtsi32_si64 (__i);
}
#ifdef __x86_64__
/* Convert I to a __m64 object. */
static __inline __m64
@ -80,6 +92,12 @@ _mm_cvtsi64_si32 (__m64 __i)
return __tmp;
}
static __inline int
_m_to_int (__m64 __i)
{
return _mm_cvtsi64_si32 (__i);
}
#ifdef __x86_64__
/* Convert the lower 32 bits of the __m64 object into an integer. */
static __inline long long
@ -98,6 +116,12 @@ _mm_packs_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_packsswb ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_packsswb (__m64 __m1, __m64 __m2)
{
return _mm_packs_pi16 (__m1, __m2);
}
/* Pack the two 32-bit values from M1 in to the lower two 16-bit values of
the result, and the two 32-bit values from M2 into the upper two 16-bit
values of the result, all with signed saturation. */
@ -107,6 +131,12 @@ _mm_packs_pi32 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_packssdw ((__v2si)__m1, (__v2si)__m2);
}
static __inline __m64
_m_packssdw (__m64 __m1, __m64 __m2)
{
return _mm_packs_pi32 (__m1, __m2);
}
/* Pack the four 16-bit values from M1 into the lower four 8-bit values of
the result, and the four 16-bit values from M2 into the upper four 8-bit
values of the result, all with unsigned saturation. */
@ -116,6 +146,12 @@ _mm_packs_pu16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_packuswb ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_packuswb (__m64 __m1, __m64 __m2)
{
return _mm_packs_pu16 (__m1, __m2);
}
/* Interleave the four 8-bit values from the high half of M1 with the four
8-bit values from the high half of M2. */
static __inline __m64
@ -124,6 +160,12 @@ _mm_unpackhi_pi8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_punpckhbw ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_punpckhbw (__m64 __m1, __m64 __m2)
{
return _mm_unpackhi_pi8 (__m1, __m2);
}
/* Interleave the two 16-bit values from the high half of M1 with the two
16-bit values from the high half of M2. */
static __inline __m64
@ -132,6 +174,12 @@ _mm_unpackhi_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_punpckhwd ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_punpckhwd (__m64 __m1, __m64 __m2)
{
return _mm_unpackhi_pi16 (__m1, __m2);
}
/* Interleave the 32-bit value from the high half of M1 with the 32-bit
value from the high half of M2. */
static __inline __m64
@ -140,6 +188,12 @@ _mm_unpackhi_pi32 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_punpckhdq ((__v2si)__m1, (__v2si)__m2);
}
static __inline __m64
_m_punpckhdq (__m64 __m1, __m64 __m2)
{
return _mm_unpackhi_pi32 (__m1, __m2);
}
/* Interleave the four 8-bit values from the low half of M1 with the four
8-bit values from the low half of M2. */
static __inline __m64
@ -148,6 +202,12 @@ _mm_unpacklo_pi8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_punpcklbw ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_punpcklbw (__m64 __m1, __m64 __m2)
{
return _mm_unpacklo_pi8 (__m1, __m2);
}
/* Interleave the two 16-bit values from the low half of M1 with the two
16-bit values from the low half of M2. */
static __inline __m64
@ -156,6 +216,12 @@ _mm_unpacklo_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_punpcklwd ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_punpcklwd (__m64 __m1, __m64 __m2)
{
return _mm_unpacklo_pi16 (__m1, __m2);
}
/* Interleave the 32-bit value from the low half of M1 with the 32-bit
value from the low half of M2. */
static __inline __m64
@ -164,6 +230,12 @@ _mm_unpacklo_pi32 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_punpckldq ((__v2si)__m1, (__v2si)__m2);
}
static __inline __m64
_m_punpckldq (__m64 __m1, __m64 __m2)
{
return _mm_unpacklo_pi32 (__m1, __m2);
}
/* Add the 8-bit values in M1 to the 8-bit values in M2. */
static __inline __m64
_mm_add_pi8 (__m64 __m1, __m64 __m2)
@ -171,6 +243,12 @@ _mm_add_pi8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_paddb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_paddb (__m64 __m1, __m64 __m2)
{
return _mm_add_pi8 (__m1, __m2);
}
/* Add the 16-bit values in M1 to the 16-bit values in M2. */
static __inline __m64
_mm_add_pi16 (__m64 __m1, __m64 __m2)
@ -178,6 +256,12 @@ _mm_add_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_paddw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_paddw (__m64 __m1, __m64 __m2)
{
return _mm_add_pi16 (__m1, __m2);
}
/* Add the 32-bit values in M1 to the 32-bit values in M2. */
static __inline __m64
_mm_add_pi32 (__m64 __m1, __m64 __m2)
@ -185,6 +269,12 @@ _mm_add_pi32 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_paddd ((__v2si)__m1, (__v2si)__m2);
}
static __inline __m64
_m_paddd (__m64 __m1, __m64 __m2)
{
return _mm_add_pi32 (__m1, __m2);
}
/* Add the 64-bit values in M1 to the 64-bit values in M2. */
static __inline __m64
_mm_add_si64 (__m64 __m1, __m64 __m2)
@ -200,6 +290,12 @@ _mm_adds_pi8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_paddsb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_paddsb (__m64 __m1, __m64 __m2)
{
return _mm_adds_pi8 (__m1, __m2);
}
/* Add the 16-bit values in M1 to the 16-bit values in M2 using signed
saturated arithmetic. */
static __inline __m64
@ -208,6 +304,12 @@ _mm_adds_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_paddsw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_paddsw (__m64 __m1, __m64 __m2)
{
return _mm_adds_pi16 (__m1, __m2);
}
/* Add the 8-bit values in M1 to the 8-bit values in M2 using unsigned
saturated arithmetic. */
static __inline __m64
@ -216,6 +318,12 @@ _mm_adds_pu8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_paddusb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_paddusb (__m64 __m1, __m64 __m2)
{
return _mm_adds_pu8 (__m1, __m2);
}
/* Add the 16-bit values in M1 to the 16-bit values in M2 using unsigned
saturated arithmetic. */
static __inline __m64
@ -224,6 +332,12 @@ _mm_adds_pu16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_paddusw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_paddusw (__m64 __m1, __m64 __m2)
{
return _mm_adds_pu16 (__m1, __m2);
}
/* Subtract the 8-bit values in M2 from the 8-bit values in M1. */
static __inline __m64
_mm_sub_pi8 (__m64 __m1, __m64 __m2)
@ -231,6 +345,12 @@ _mm_sub_pi8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_psubb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_psubb (__m64 __m1, __m64 __m2)
{
return _mm_sub_pi8 (__m1, __m2);
}
/* Subtract the 16-bit values in M2 from the 16-bit values in M1. */
static __inline __m64
_mm_sub_pi16 (__m64 __m1, __m64 __m2)
@ -238,6 +358,12 @@ _mm_sub_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_psubw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_psubw (__m64 __m1, __m64 __m2)
{
return _mm_sub_pi16 (__m1, __m2);
}
/* Subtract the 32-bit values in M2 from the 32-bit values in M1. */
static __inline __m64
_mm_sub_pi32 (__m64 __m1, __m64 __m2)
@ -245,6 +371,12 @@ _mm_sub_pi32 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_psubd ((__v2si)__m1, (__v2si)__m2);
}
static __inline __m64
_m_psubd (__m64 __m1, __m64 __m2)
{
return _mm_sub_pi32 (__m1, __m2);
}
/* Add the 64-bit values in M1 to the 64-bit values in M2. */
static __inline __m64
_mm_sub_si64 (__m64 __m1, __m64 __m2)
@ -260,6 +392,12 @@ _mm_subs_pi8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_psubsb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_psubsb (__m64 __m1, __m64 __m2)
{
return _mm_subs_pi8 (__m1, __m2);
}
/* Subtract the 16-bit values in M2 from the 16-bit values in M1 using
signed saturating arithmetic. */
static __inline __m64
@ -268,6 +406,12 @@ _mm_subs_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_psubsw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_psubsw (__m64 __m1, __m64 __m2)
{
return _mm_subs_pi16 (__m1, __m2);
}
/* Subtract the 8-bit values in M2 from the 8-bit values in M1 using
unsigned saturating arithmetic. */
static __inline __m64
@ -276,6 +420,12 @@ _mm_subs_pu8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_psubusb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_psubusb (__m64 __m1, __m64 __m2)
{
return _mm_subs_pu8 (__m1, __m2);
}
/* Subtract the 16-bit values in M2 from the 16-bit values in M1 using
unsigned saturating arithmetic. */
static __inline __m64
@ -284,6 +434,12 @@ _mm_subs_pu16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_psubusw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_psubusw (__m64 __m1, __m64 __m2)
{
return _mm_subs_pu16 (__m1, __m2);
}
/* Multiply four 16-bit values in M1 by four 16-bit values in M2 producing
four 32-bit intermediate results, which are then summed by pairs to
produce two 32-bit results. */
@ -293,6 +449,12 @@ _mm_madd_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pmaddwd ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_pmaddwd (__m64 __m1, __m64 __m2)
{
return _mm_madd_pi16 (__m1, __m2);
}
/* Multiply four signed 16-bit values in M1 by four signed 16-bit values in
M2 and produce the high 16 bits of the 32-bit results. */
static __inline __m64
@ -301,6 +463,12 @@ _mm_mulhi_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pmulhw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_pmulhw (__m64 __m1, __m64 __m2)
{
return _mm_mulhi_pi16 (__m1, __m2);
}
/* Multiply four 16-bit values in M1 by four 16-bit values in M2 and produce
the low 16 bits of the results. */
static __inline __m64
@ -309,6 +477,12 @@ _mm_mullo_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pmullw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_pmullw (__m64 __m1, __m64 __m2)
{
return _mm_mullo_pi16 (__m1, __m2);
}
/* Shift four 16-bit values in M left by COUNT. */
static __inline __m64
_mm_sll_pi16 (__m64 __m, __m64 __count)
@ -316,12 +490,24 @@ _mm_sll_pi16 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_psllw ((__v4hi)__m, (long long)__count);
}
static __inline __m64
_m_psllw (__m64 __m, __m64 __count)
{
return _mm_sll_pi16 (__m, __count);
}
static __inline __m64
_mm_slli_pi16 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_psllw ((__v4hi)__m, __count);
}
static __inline __m64
_m_psllwi (__m64 __m, int __count)
{
return _mm_slli_pi16 (__m, __count);
}
/* Shift two 32-bit values in M left by COUNT. */
static __inline __m64
_mm_sll_pi32 (__m64 __m, __m64 __count)
@ -329,12 +515,24 @@ _mm_sll_pi32 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_pslld ((__v2si)__m, (long long)__count);
}
static __inline __m64
_m_pslld (__m64 __m, __m64 __count)
{
return _mm_sll_pi32 (__m, __count);
}
static __inline __m64
_mm_slli_pi32 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_pslld ((__v2si)__m, __count);
}
static __inline __m64
_m_pslldi (__m64 __m, int __count)
{
return _mm_slli_pi32 (__m, __count);
}
/* Shift the 64-bit value in M left by COUNT. */
static __inline __m64
_mm_sll_si64 (__m64 __m, __m64 __count)
@ -342,12 +540,24 @@ _mm_sll_si64 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_psllq ((long long)__m, (long long)__count);
}
static __inline __m64
_m_psllq (__m64 __m, __m64 __count)
{
return _mm_sll_si64 (__m, __count);
}
static __inline __m64
_mm_slli_si64 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_psllq ((long long)__m, (long long)__count);
}
static __inline __m64
_m_psllqi (__m64 __m, int __count)
{
return _mm_slli_si64 (__m, __count);
}
/* Shift four 16-bit values in M right by COUNT; shift in the sign bit. */
static __inline __m64
_mm_sra_pi16 (__m64 __m, __m64 __count)
@ -355,12 +565,24 @@ _mm_sra_pi16 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_psraw ((__v4hi)__m, (long long)__count);
}
static __inline __m64
_m_psraw (__m64 __m, __m64 __count)
{
return _mm_sra_pi16 (__m, __count);
}
static __inline __m64
_mm_srai_pi16 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_psraw ((__v4hi)__m, __count);
}
static __inline __m64
_m_psrawi (__m64 __m, int __count)
{
return _mm_srai_pi16 (__m, __count);
}
/* Shift two 32-bit values in M right by COUNT; shift in the sign bit. */
static __inline __m64
_mm_sra_pi32 (__m64 __m, __m64 __count)
@ -368,12 +590,24 @@ _mm_sra_pi32 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_psrad ((__v2si)__m, (long long)__count);
}
static __inline __m64
_m_psrad (__m64 __m, __m64 __count)
{
return _mm_sra_pi32 (__m, __count);
}
static __inline __m64
_mm_srai_pi32 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_psrad ((__v2si)__m, __count);
}
static __inline __m64
_m_psradi (__m64 __m, int __count)
{
return _mm_srai_pi32 (__m, __count);
}
/* Shift four 16-bit values in M right by COUNT; shift in zeros. */
static __inline __m64
_mm_srl_pi16 (__m64 __m, __m64 __count)
@ -381,12 +615,24 @@ _mm_srl_pi16 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_psrlw ((__v4hi)__m, (long long)__count);
}
static __inline __m64
_m_psrlw (__m64 __m, __m64 __count)
{
return _mm_srl_pi16 (__m, __count);
}
static __inline __m64
_mm_srli_pi16 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_psrlw ((__v4hi)__m, __count);
}
static __inline __m64
_m_psrlwi (__m64 __m, int __count)
{
return _mm_srli_pi16 (__m, __count);
}
/* Shift two 32-bit values in M right by COUNT; shift in zeros. */
static __inline __m64
_mm_srl_pi32 (__m64 __m, __m64 __count)
@ -394,12 +640,24 @@ _mm_srl_pi32 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_psrld ((__v2si)__m, (long long)__count);
}
static __inline __m64
_m_psrld (__m64 __m, __m64 __count)
{
return _mm_srl_pi32 (__m, __count);
}
static __inline __m64
_mm_srli_pi32 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_psrld ((__v2si)__m, __count);
}
static __inline __m64
_m_psrldi (__m64 __m, int __count)
{
return _mm_srli_pi32 (__m, __count);
}
/* Shift the 64-bit value in M left by COUNT; shift in zeros. */
static __inline __m64
_mm_srl_si64 (__m64 __m, __m64 __count)
@ -407,12 +665,24 @@ _mm_srl_si64 (__m64 __m, __m64 __count)
return (__m64) __builtin_ia32_psrlq ((long long)__m, (long long)__count);
}
static __inline __m64
_m_psrlq (__m64 __m, __m64 __count)
{
return _mm_srl_si64 (__m, __count);
}
static __inline __m64
_mm_srli_si64 (__m64 __m, int __count)
{
return (__m64) __builtin_ia32_psrlq ((long long)__m, (long long)__count);
}
static __inline __m64
_m_psrlqi (__m64 __m, int __count)
{
return _mm_srli_si64 (__m, __count);
}
/* Bit-wise AND the 64-bit values in M1 and M2. */
static __inline __m64
_mm_and_si64 (__m64 __m1, __m64 __m2)
@ -420,6 +690,12 @@ _mm_and_si64 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pand ((long long)__m1, (long long)__m2);
}
static __inline __m64
_m_pand (__m64 __m1, __m64 __m2)
{
return _mm_and_si64 (__m1, __m2);
}
/* Bit-wise complement the 64-bit value in M1 and bit-wise AND it with the
64-bit value in M2. */
static __inline __m64
@ -428,6 +704,12 @@ _mm_andnot_si64 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pandn ((long long)__m1, (long long)__m2);
}
static __inline __m64
_m_pandn (__m64 __m1, __m64 __m2)
{
return _mm_andnot_si64 (__m1, __m2);
}
/* Bit-wise inclusive OR the 64-bit values in M1 and M2. */
static __inline __m64
_mm_or_si64 (__m64 __m1, __m64 __m2)
@ -435,6 +717,12 @@ _mm_or_si64 (__m64 __m1, __m64 __m2)
return (__m64)__builtin_ia32_por ((long long)__m1, (long long)__m2);
}
static __inline __m64
_m_por (__m64 __m1, __m64 __m2)
{
return _mm_or_si64 (__m1, __m2);
}
/* Bit-wise exclusive OR the 64-bit values in M1 and M2. */
static __inline __m64
_mm_xor_si64 (__m64 __m1, __m64 __m2)
@ -442,6 +730,12 @@ _mm_xor_si64 (__m64 __m1, __m64 __m2)
return (__m64)__builtin_ia32_pxor ((long long)__m1, (long long)__m2);
}
static __inline __m64
_m_pxor (__m64 __m1, __m64 __m2)
{
return _mm_xor_si64 (__m1, __m2);
}
/* Compare eight 8-bit values. The result of the comparison is 0xFF if the
test is true and zero if false. */
static __inline __m64
@ -450,12 +744,24 @@ _mm_cmpeq_pi8 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pcmpeqb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_pcmpeqb (__m64 __m1, __m64 __m2)
{
return _mm_cmpeq_pi8 (__m1, __m2);
}
static __inline __m64
_mm_cmpgt_pi8 (__m64 __m1, __m64 __m2)
{
return (__m64) __builtin_ia32_pcmpgtb ((__v8qi)__m1, (__v8qi)__m2);
}
static __inline __m64
_m_pcmpgtb (__m64 __m1, __m64 __m2)
{
return _mm_cmpgt_pi8 (__m1, __m2);
}
/* Compare four 16-bit values. The result of the comparison is 0xFFFF if
the test is true and zero if false. */
static __inline __m64
@ -464,12 +770,24 @@ _mm_cmpeq_pi16 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pcmpeqw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_pcmpeqw (__m64 __m1, __m64 __m2)
{
return _mm_cmpeq_pi16 (__m1, __m2);
}
static __inline __m64
_mm_cmpgt_pi16 (__m64 __m1, __m64 __m2)
{
return (__m64) __builtin_ia32_pcmpgtw ((__v4hi)__m1, (__v4hi)__m2);
}
static __inline __m64
_m_pcmpgtw (__m64 __m1, __m64 __m2)
{
return _mm_cmpgt_pi16 (__m1, __m2);
}
/* Compare two 32-bit values. The result of the comparison is 0xFFFFFFFF if
the test is true and zero if false. */
static __inline __m64
@ -478,12 +796,24 @@ _mm_cmpeq_pi32 (__m64 __m1, __m64 __m2)
return (__m64) __builtin_ia32_pcmpeqd ((__v2si)__m1, (__v2si)__m2);
}
static __inline __m64
_m_pcmpeqd (__m64 __m1, __m64 __m2)
{
return _mm_cmpeq_pi32 (__m1, __m2);
}
static __inline __m64
_mm_cmpgt_pi32 (__m64 __m1, __m64 __m2)
{
return (__m64) __builtin_ia32_pcmpgtd ((__v2si)__m1, (__v2si)__m2);
}
static __inline __m64
_m_pcmpgtd (__m64 __m1, __m64 __m2)
{
return _mm_cmpgt_pi32 (__m1, __m2);
}
/* Creates a 64-bit zero. */
static __inline __m64
_mm_setzero_si64 (void)
@ -583,62 +913,5 @@ _mm_set1_pi8 (char __b)
return _mm_set1_pi32 (__i);
}
/* Alternate intrinsic name definitions. */
#define _m_empty _mm_empty
#define _m_from_int _mm_cvtsi32_si64
#define _m_to_int _mm_cvtsi64_si32
#define _m_packsswb _mm_packs_pi16
#define _m_packssdw _mm_packs_pi32
#define _m_packuswb _mm_packs_pu16
#define _m_punpckhbw _mm_unpackhi_pi8
#define _m_punpckhwd _mm_unpackhi_pi16
#define _m_punpckhdq _mm_unpackhi_pi32
#define _m_punpcklbw _mm_unpacklo_pi8
#define _m_punpcklwd _mm_unpacklo_pi16
#define _m_punpckldq _mm_unpacklo_pi32
#define _m_paddb _mm_add_pi8
#define _m_paddw _mm_add_pi16
#define _m_paddd _mm_add_pi32
#define _m_paddsb _mm_adds_pi8
#define _m_paddsw _mm_adds_pi16
#define _m_paddusb _mm_adds_pu8
#define _m_paddusw _mm_adds_pu16
#define _m_psubb _mm_sub_pi8
#define _m_psubw _mm_sub_pi16
#define _m_psubd _mm_sub_pi32
#define _m_psubsb _mm_subs_pi8
#define _m_psubsw _mm_subs_pi16
#define _m_psubusb _mm_subs_pu8
#define _m_psubusw _mm_subs_pu16
#define _m_pmaddwd _mm_madd_pi16
#define _m_pmulhw _mm_mulhi_pi16
#define _m_pmullw _mm_mullo_pi16
#define _m_psllw _mm_sll_pi16
#define _m_psllwi _mm_slli_pi16
#define _m_pslld _mm_sll_pi32
#define _m_pslldi _mm_slli_pi32
#define _m_psllq _mm_sll_si64
#define _m_psllqi _mm_slli_si64
#define _m_psraw _mm_sra_pi16
#define _m_psrawi _mm_srai_pi16
#define _m_psrad _mm_sra_pi32
#define _m_psradi _mm_srai_pi32
#define _m_psrlw _mm_srl_pi16
#define _m_psrlwi _mm_srli_pi16
#define _m_psrld _mm_srl_pi32
#define _m_psrldi _mm_srli_pi32
#define _m_psrlq _mm_srl_si64
#define _m_psrlqi _mm_srli_si64
#define _m_pand _mm_and_si64
#define _m_pandn _mm_andnot_si64
#define _m_por _mm_or_si64
#define _m_pxor _mm_xor_si64
#define _m_pcmpeqb _mm_cmpeq_pi8
#define _m_pcmpeqw _mm_cmpeq_pi16
#define _m_pcmpeqd _mm_cmpeq_pi32
#define _m_pcmpgtb _mm_cmpgt_pi8
#define _m_pcmpgtw _mm_cmpgt_pi16
#define _m_pcmpgtd _mm_cmpgt_pi32
#endif /* __MMX__ */
#endif /* _MMINTRIN_H_INCLUDED */

View File

@ -475,6 +475,12 @@ _mm_cvtss_si32 (__m128 __A)
return __builtin_ia32_cvtss2si ((__v4sf) __A);
}
static __inline int
_mm_cvt_ss2si (__m128 __A)
{
return _mm_cvtss_si32 (__A);
}
#ifdef __x86_64__
/* Convert the lower SPFP value to a 32-bit integer according to the current
rounding mode. */
@ -493,6 +499,12 @@ _mm_cvtps_pi32 (__m128 __A)
return (__m64) __builtin_ia32_cvtps2pi ((__v4sf) __A);
}
static __inline __m64
_mm_cvt_ps2pi (__m128 __A)
{
return _mm_cvtps_pi32 (__A);
}
/* Truncate the lower SPFP value to a 32-bit integer. */
static __inline int
_mm_cvttss_si32 (__m128 __A)
@ -500,6 +512,12 @@ _mm_cvttss_si32 (__m128 __A)
return __builtin_ia32_cvttss2si ((__v4sf) __A);
}
static __inline int
_mm_cvtt_ss2si (__m128 __A)
{
return _mm_cvttss_si32 (__A);
}
#ifdef __x86_64__
/* Truncate the lower SPFP value to a 32-bit integer. */
static __inline long long
@ -517,6 +535,12 @@ _mm_cvttps_pi32 (__m128 __A)
return (__m64) __builtin_ia32_cvttps2pi ((__v4sf) __A);
}
static __inline __m64
_mm_cvtt_ps2pi (__m128 __A)
{
return _mm_cvttps_pi32 (__A);
}
/* Convert B to a SPFP value and insert it as element zero in A. */
static __inline __m128
_mm_cvtsi32_ss (__m128 __A, int __B)
@ -524,6 +548,12 @@ _mm_cvtsi32_ss (__m128 __A, int __B)
return (__m128) __builtin_ia32_cvtsi2ss ((__v4sf) __A, __B);
}
static __inline __m128
_mm_cvt_si2ss (__m128 __A, int __B)
{
return _mm_cvtsi32_ss (__A, __B);
}
#ifdef __x86_64__
/* Convert B to a SPFP value and insert it as element zero in A. */
static __inline __m128
@ -541,6 +571,12 @@ _mm_cvtpi32_ps (__m128 __A, __m64 __B)
return (__m128) __builtin_ia32_cvtpi2ps ((__v4sf) __A, (__v2si)__B);
}
static __inline __m128
_mm_cvt_pi2ps (__m128 __A, __m64 __B)
{
return _mm_cvtpi32_ps (__A, __B);
}
/* Convert the four signed 16-bit values in A to SPFP form. */
static __inline __m128
_mm_cvtpi16_ps (__m64 __A)
@ -942,9 +978,16 @@ _mm_extract_pi16 (__m64 __A, int __N)
{
return __builtin_ia32_pextrw ((__v4hi)__A, __N);
}
static __inline int
_m_pextrw (__m64 __A, int __N)
{
return _mm_extract_pi16 (__A, __N);
}
#else
#define _mm_extract_pi16(A, N) \
__builtin_ia32_pextrw ((__v4hi)(A), (N))
#define _m_pextrw(A, N) _mm_extract_pi16((A), (N))
#endif
/* Inserts word D into one of four words of A. The selector N must be
@ -955,9 +998,16 @@ _mm_insert_pi16 (__m64 __A, int __D, int __N)
{
return (__m64)__builtin_ia32_pinsrw ((__v4hi)__A, __D, __N);
}
static __inline __m64
_m_pinsrw (__m64 __A, int __D, int __N)
{
return _mm_insert_pi16 (__A, __D, __N);
}
#else
#define _mm_insert_pi16(A, D, N) \
((__m64) __builtin_ia32_pinsrw ((__v4hi)(A), (D), (N)))
#define _m_pinsrw(A, D, N) _mm_insert_pi16((A), (D), (N))
#endif
/* Compute the element-wise maximum of signed 16-bit values. */
@ -967,6 +1017,12 @@ _mm_max_pi16 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_pmaxsw ((__v4hi)__A, (__v4hi)__B);
}
static __inline __m64
_m_pmaxsw (__m64 __A, __m64 __B)
{
return _mm_max_pi16 (__A, __B);
}
/* Compute the element-wise maximum of unsigned 8-bit values. */
static __inline __m64
_mm_max_pu8 (__m64 __A, __m64 __B)
@ -974,6 +1030,12 @@ _mm_max_pu8 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_pmaxub ((__v8qi)__A, (__v8qi)__B);
}
static __inline __m64
_m_pmaxub (__m64 __A, __m64 __B)
{
return _mm_max_pu8 (__A, __B);
}
/* Compute the element-wise minimum of signed 16-bit values. */
static __inline __m64
_mm_min_pi16 (__m64 __A, __m64 __B)
@ -981,6 +1043,12 @@ _mm_min_pi16 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_pminsw ((__v4hi)__A, (__v4hi)__B);
}
static __inline __m64
_m_pminsw (__m64 __A, __m64 __B)
{
return _mm_min_pi16 (__A, __B);
}
/* Compute the element-wise minimum of unsigned 8-bit values. */
static __inline __m64
_mm_min_pu8 (__m64 __A, __m64 __B)
@ -988,6 +1056,12 @@ _mm_min_pu8 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_pminub ((__v8qi)__A, (__v8qi)__B);
}
static __inline __m64
_m_pminub (__m64 __A, __m64 __B)
{
return _mm_min_pu8 (__A, __B);
}
/* Create an 8-bit mask of the signs of 8-bit values. */
static __inline int
_mm_movemask_pi8 (__m64 __A)
@ -995,6 +1069,12 @@ _mm_movemask_pi8 (__m64 __A)
return __builtin_ia32_pmovmskb ((__v8qi)__A);
}
static __inline int
_m_pmovmskb (__m64 __A)
{
return _mm_movemask_pi8 (__A);
}
/* Multiply four unsigned 16-bit values in A by four unsigned 16-bit values
in B and produce the high 16 bits of the 32-bit results. */
static __inline __m64
@ -1003,6 +1083,12 @@ _mm_mulhi_pu16 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_pmulhuw ((__v4hi)__A, (__v4hi)__B);
}
static __inline __m64
_m_pmulhuw (__m64 __A, __m64 __B)
{
return _mm_mulhi_pu16 (__A, __B);
}
/* Return a combination of the four 16-bit values in A. The selector
must be an immediate. */
#if 0
@ -1011,9 +1097,16 @@ _mm_shuffle_pi16 (__m64 __A, int __N)
{
return (__m64) __builtin_ia32_pshufw ((__v4hi)__A, __N);
}
static __inline __m64
_m_pshufw (__m64 __A, int __N)
{
return _mm_shuffle_pi16 (__A, __N);
}
#else
#define _mm_shuffle_pi16(A, N) \
((__m64) __builtin_ia32_pshufw ((__v4hi)(A), (N)))
#define _m_pshufw(A, N) _mm_shuffle_pi16 ((A), (N))
#endif
/* Conditionally store byte elements of A into P. The high bit of each
@ -1025,6 +1118,12 @@ _mm_maskmove_si64 (__m64 __A, __m64 __N, char *__P)
__builtin_ia32_maskmovq ((__v8qi)__A, (__v8qi)__N, __P);
}
static __inline void
_m_maskmovq (__m64 __A, __m64 __N, char *__P)
{
_mm_maskmove_si64 (__A, __N, __P);
}
/* Compute the rounded averages of the unsigned 8-bit values in A and B. */
static __inline __m64
_mm_avg_pu8 (__m64 __A, __m64 __B)
@ -1032,6 +1131,12 @@ _mm_avg_pu8 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_pavgb ((__v8qi)__A, (__v8qi)__B);
}
static __inline __m64
_m_pavgb (__m64 __A, __m64 __B)
{
return _mm_avg_pu8 (__A, __B);
}
/* Compute the rounded averages of the unsigned 16-bit values in A and B. */
static __inline __m64
_mm_avg_pu16 (__m64 __A, __m64 __B)
@ -1039,6 +1144,12 @@ _mm_avg_pu16 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_pavgw ((__v4hi)__A, (__v4hi)__B);
}
static __inline __m64
_m_pavgw (__m64 __A, __m64 __B)
{
return _mm_avg_pu16 (__A, __B);
}
/* Compute the sum of the absolute differences of the unsigned 8-bit
values in A and B. Return the value in the lower 16-bit word; the
upper words are cleared. */
@ -1048,6 +1159,12 @@ _mm_sad_pu8 (__m64 __A, __m64 __B)
return (__m64) __builtin_ia32_psadbw ((__v8qi)__A, (__v8qi)__B);
}
static __inline __m64
_m_psadbw (__m64 __A, __m64 __B)
{
return _mm_sad_pu8 (__A, __B);
}
/* Loads one cache line from address P to a location "closer" to the
processor. The selector I specifies the type of prefetch operation. */
#if 0
@ -1106,30 +1223,6 @@ do { \
(row3) = __builtin_ia32_shufps (__t2, __t3, 0xDD); \
} while (0)
/* Alternate intrinsic name definitions. */
#define _mm_cvt_ss2si _mm_cvtss_si32
#define _mm_cvt_ps2pi _mm_cvtps_pi32
#define _mm_cvtt_ss2si _mm_cvttss_si32
#define _mm_cvtt_ps2pi _mm_cvttps_pi32
#define _mm_cvt_si2ss _mm_cvtsi32_ss
#define _mm_cvt_pi2ps _mm_cvtpi32_ps
#define _m_pextrw _mm_extract_pi16
#define _m_pinsrw _mm_insert_pi16
#define _m_pmaxsw _mm_max_pi16
#define _m_pmaxub _mm_max_pu8
#define _m_pminsw _mm_min_pi16
#define _m_pminub _mm_min_pu8
#define _m_pmovmskb _mm_movemask_pi8
#define _m_pmulhuw _mm_mulhi_pu16
#define _m_pshufw _mm_shuffle_pi16
#define _m_maskmovq _mm_maskmove_si64
#define _m_pavgb _mm_avg_pu8
#define _m_pavgw _mm_avg_pu16
#define _m_psadbw _mm_sad_pu8
#define _mm_set_ps1 _mm_set1_ps
#define _mm_load_ps1 _mm_load1_ps
#define _mm_store_ps1 _mm_store1_ps
/* For backward source compatibility. */
#include <emmintrin.h>