r272121 - Add doxygen comments to xmmintrin.h's intrinsics.

Ekaterina Romanova via cfe-commits cfe-commits at lists.llvm.org
Wed Jun 8 00:34:32 PDT 2016


Author: kromanova
Date: Wed Jun  8 02:34:31 2016
New Revision: 272121

URL: http://llvm.org/viewvc/llvm-project?rev=272121&view=rev
Log:
Add doxygen comments to xmmintrin.h's intrinsics.
Only half of the intrinsics in this file is documented here. The patch for the o
ther half will be sent out later.

The doxygen comments are automatically generated based on Sony's intrinsics docu
ment.

I got an OK from Eric Christopher to commit doxygen comments without prior code
review upstream.


Modified:
    cfe/trunk/lib/Headers/xmmintrin.h

Modified: cfe/trunk/lib/Headers/xmmintrin.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/xmmintrin.h?rev=272121&r1=272120&r2=272121&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/xmmintrin.h (original)
+++ cfe/trunk/lib/Headers/xmmintrin.h Wed Jun  8 02:34:31 2016
@@ -1406,12 +1406,39 @@ _mm_cvttps_pi32(__m128 __a)
   return (__m64)__builtin_ia32_cvttps2pi((__v4sf)__a);
 }
 
+/// \brief Converts two low-order float values in a 128-bit vector of [4 x
+///    float] into a 64-bit vector of [2 x i32], truncating the result when it
+///    is inexact.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTTPS2PI instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+/// \returns A 64-bit integer vector containing the converted values.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_cvtt_ps2pi(__m128 __a)
 {
   return _mm_cvttps_pi32(__a);
 }
 
+/// \brief Converts a 32-bit signed integer value into a floating point value
+///    and writes it to the lower 32 bits of the destination. The remaining
+///    higher order elements of the destination vector are copied from the
+///    corresponding elements in the first operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VCVTSI2SS / CVTSI2SS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+/// \param __b
+///    A 32-bit signed integer operand containing the value to be converted.
+/// \returns A 128-bit vector of [4 x float] whose lower 32 bits contain the
+///    converted value of the second operand. The upper 96 bits are copied from
+///    the upper 96 bits of the first operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtsi32_ss(__m128 __a, int __b)
 {
@@ -1419,6 +1446,22 @@ _mm_cvtsi32_ss(__m128 __a, int __b)
   return __a;
 }
 
+/// \brief Converts a 32-bit signed integer value into a floating point value
+///    and writes it to the lower 32 bits of the destination. The remaining
+///    higher order elements of the destination are copied from the
+///    corresponding elements in the first operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VCVTSI2SS / CVTSI2SS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+/// \param __b
+///    A 32-bit signed integer operand containing the value to be converted.
+/// \returns A 128-bit vector of [4 x float] whose lower 32 bits contain the
+///    converted value of the second operand. The upper 96 bits are copied from
+///    the upper 96 bits of the first operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvt_si2ss(__m128 __a, int __b)
 {
@@ -1427,6 +1470,22 @@ _mm_cvt_si2ss(__m128 __a, int __b)
 
 #ifdef __x86_64__
 
+/// \brief Converts a 64-bit signed integer value into a floating point value
+///    and writes it to the lower 32 bits of the destination. The remaining
+///    higher order elements of the destination are copied from the
+///    corresponding elements in the first operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VCVTSI2SS / CVTSI2SS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+/// \param __b
+///    A 64-bit signed integer operand containing the value to be converted.
+/// \returns A 128-bit vector of [4 x float] whose lower 32 bits contain the
+///    converted value of the second operand. The upper 96 bits are copied from
+///    the upper 96 bits of the first operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtsi64_ss(__m128 __a, long long __b)
 {
@@ -1436,24 +1495,84 @@ _mm_cvtsi64_ss(__m128 __a, long long __b
 
 #endif
 
+/// \brief Converts two elements of a 64-bit vector of [2 x i32] into two
+///    floating point values and writes them to the lower 64-bits of the
+///    destination. The remaining higher order elements of the destination are
+///    copied from the corresponding elements in the first operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPI2PS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+/// \param __b
+///    A 64-bit vector of [2 x i32]. The elements in this vector are converted
+///    and written to the corresponding low-order elements in the destination.
+/// \returns A 128-bit vector of [4 x float] whose lower 64 bits contain the
+///    converted value of the second operand. The upper 64 bits are copied from
+///    the upper 64 bits of the first operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtpi32_ps(__m128 __a, __m64 __b)
 {
   return __builtin_ia32_cvtpi2ps((__v4sf)__a, (__v2si)__b);
 }
 
+/// \brief Converts two elements of a 64-bit vector of [2 x i32] into two
+///    floating point values and writes them to the lower 64-bits of the
+///    destination. The remaining higher order elements of the destination are
+///    copied from the corresponding elements in the first operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPI2PS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+/// \param __b
+///    A 64-bit vector of [2 x i32]. The elements in this vector are converted
+///    and written to the corresponding low-order elements in the destination.
+/// \returns A 128-bit vector of [4 x float] whose lower 64 bits contain the
+///    converted value from the second operand. The upper 64 bits are copied
+///    from the upper 64 bits of the first operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvt_pi2ps(__m128 __a, __m64 __b)
 {
   return _mm_cvtpi32_ps(__a, __b);
 }
 
+/// \brief Extracts a float value contained in the lower 32 bits of a vector of
+///    [4 x float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVSS / MOVSS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float]. The lower 32 bits of this operand are
+///    used in the extraction.
+/// \returns A 32-bit float containing the extracted value.
 static __inline__ float __DEFAULT_FN_ATTRS
 _mm_cvtss_f32(__m128 __a)
 {
   return __a[0];
 }
 
+/// \brief Loads two packed float values from the address __p into the
+///     high-order bits of a 128-bit vector of [4 x float]. The low-order bits
+///     are copied from the low-order bits of the first operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVHPD / MOVHPD instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float]. Bits [63:0] are written to bits [63:0]
+///    of the destination.
+/// \param __p
+///    A pointer to two packed float values. Bits [63:0] are written to bits
+///    [127:64] of the destination.
+/// \returns A 128-bit vector of [4 x float] containing the moved values.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_loadh_pi(__m128 __a, const __m64 *__p)
 {
@@ -1466,6 +1585,21 @@ _mm_loadh_pi(__m128 __a, const __m64 *__
   return __builtin_shufflevector(__a, __bb, 0, 1, 4, 5);
 }
 
+/// \brief Loads two packed float values from the address __p into the low-order
+///    bits of a 128-bit vector of [4 x float]. The high-order bits are copied
+///    from the high-order bits of the first operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVLPD / MOVLPD instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float]. Bits [127:64] are written to bits
+///    [127:64] of the destination.
+/// \param __p
+///    A pointer to two packed float values. Bits [63:0] are written to bits
+///    [63:0] of the destination.
+/// \returns A 128-bit vector of [4 x float] containing the moved values.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_loadl_pi(__m128 __a, const __m64 *__p)
 {
@@ -1478,6 +1612,21 @@ _mm_loadl_pi(__m128 __a, const __m64 *__
   return __builtin_shufflevector(__a, __bb, 4, 5, 2, 3);
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float]. The lower
+///    32 bits of the vector are initialized with the single-precision
+///    floating-point value loaded from a specified memory location. The upper
+///    96 bits are set to zero.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVSS / MOVSS instruction.
+///
+/// \param __p
+///    A pointer to a 32-bit memory location containing a single-precision
+///    floating-point value.
+/// \returns An initialized 128-bit floating-point vector of [4 x float]. The
+///    lower 32 bits contain the value loaded from the memory location. The
+///    upper 96 bits are set to zero.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_load_ss(const float *__p)
 {
@@ -1488,6 +1637,18 @@ _mm_load_ss(const float *__p)
   return (__m128){ __u, 0, 0, 0 };
 }
 
+/// \brief Loads a 32-bit float value and duplicates it to all four vector
+///    elements of a 128-bit vector of [4 x float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVSS / MOVSS + \c shuffling
+///    instruction.
+///
+/// \param __p
+///    A pointer to a float value to be loaded and duplicated.
+/// \returns A 128-bit vector of [4 x float] containing the loaded
+///    and duplicated values.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_load1_ps(const float *__p)
 {
@@ -1500,12 +1661,34 @@ _mm_load1_ps(const float *__p)
 
 #define        _mm_load_ps1(p) _mm_load1_ps(p)
 
+/// \brief Loads a 128-bit floating-point vector of [4 x float] from an aligned
+///    memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVAPS / MOVAPS instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location. The address of the memory
+///    location has to be 128-bit aligned.
+/// \returns A 128-bit vector of [4 x float] containing the loaded valus.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_load_ps(const float *__p)
 {
   return *(__m128*)__p;
 }
 
+/// \brief Loads a 128-bit floating-point vector of [4 x float] from an
+///    unaligned memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVUPS / MOVUPS instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location. The address of the memory
+///    location does not have to be aligned.
+/// \returns A 128-bit vector of [4 x float] containing the loaded values.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_loadu_ps(const float *__p)
 {
@@ -1515,6 +1698,19 @@ _mm_loadu_ps(const float *__p)
   return ((struct __loadu_ps*)__p)->__v;
 }
 
+/// \brief Loads four packed float values, in reverse order, from an aligned
+///    memory location to 32-bit elements in a 128-bit vector of [4 x float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVAPS / MOVAPS + \c shuffling
+///    instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location. The address of the memory
+///    location has to be 128-bit aligned.
+/// \returns A 128-bit vector of [4 x float] containing the moved values, loaded
+///    in reverse order.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_loadr_ps(const float *__p)
 {
@@ -1522,18 +1718,52 @@ _mm_loadr_ps(const float *__p)
   return __builtin_shufflevector((__v4sf)__a, (__v4sf)__a, 3, 2, 1, 0);
 }
 
+/// \brief Create a 128-bit vector of [4 x float] with undefined values.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic has no corresponding instruction.
+///
+/// \returns A 128-bit vector of [4 x float] containing undefined values.
+
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_undefined_ps()
 {
   return (__m128)__builtin_ia32_undef128();
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float]. The lower
+///    32 bits of the vector are initialized with the specified single-precision
+///    floating-point value. The upper 96 bits are set to zero.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVSS / MOVSS instruction.
+///
+/// \param __w
+///    A single-precision floating-point value used to initialize the lower 32
+///    bits of the result.
+/// \returns An initialized 128-bit floating-point vector of [4 x float]. The
+///    lower 32 bits contain the value provided in the source operand. The
+///    upper 96 bits are set to zero.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_set_ss(float __w)
 {
   return (__m128){ __w, 0, 0, 0 };
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float], with each
+///    of the four single-precision floating-point vector elements set to the
+///    specified single-precision floating-point value.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VPERMILPS / PERMILPS instruction.
+///
+/// \param __w
+///    A single-precision floating-point value used to initialize each vector
+///    element of the result.
+/// \returns An initialized 128-bit floating-point vector of [4 x float].
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_set1_ps(float __w)
 {
@@ -1541,42 +1771,139 @@ _mm_set1_ps(float __w)
 }
 
 /* Microsoft specific. */
+/// \brief Constructs a 128-bit floating-point vector of [4 x float], with each
+///    of the four single-precision floating-point vector elements set to the
+///    specified single-precision floating-point value.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VPERMILPS / PERMILPS instruction.
+///
+/// \param __w
+///    A single-precision floating-point value used to initialize each vector
+///    element of the result.
+/// \returns An initialized 128-bit floating-point vector of [4 x float].
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_set_ps1(float __w)
 {
     return _mm_set1_ps(__w);
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float]
+///    initialized with the specified single-precision floating-point values.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic is a utility function and does not correspond to a specific
+///    instruction.
+///
+/// \param __z
+///    A single-precision floating-point value used to initialize bits [127:96]
+///    of the result.
+/// \param __y
+///    A single-precision floating-point value used to initialize bits [95:64]
+///    of the result.
+/// \param __x
+///    A single-precision floating-point value used to initialize bits [63:32]
+///    of the result.
+/// \param __w
+///    A single-precision floating-point value used to initialize bits [31:0]
+///    of the result.
+/// \returns An initialized 128-bit floating-point vector of [4 x float].
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_set_ps(float __z, float __y, float __x, float __w)
 {
   return (__m128){ __w, __x, __y, __z };
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float],
+///    initialized in reverse order with the specified 32-bit single-precision
+///    float-point values.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic is a utility function and does not correspond to a specific
+///    instruction.
+///
+/// \param __z
+///    A single-precision floating-point value used to initialize bits [31:0]
+///    of the result.
+/// \param __y
+///    A single-precision floating-point value used to initialize bits [63:32]
+///    of the result.
+/// \param __x
+///    A single-precision floating-point value used to initialize bits [95:64]
+///    of the result.
+/// \param __w
+///    A single-precision floating-point value used to initialize bits [127:96]
+///    of the result.
+/// \returns An initialized 128-bit floating-point vector of [4 x float].
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_setr_ps(float __z, float __y, float __x, float __w)
 {
   return (__m128){ __z, __y, __x, __w };
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float] initialized
+///    to zero.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VXORPS / XORPS instruction.
+///
+/// \returns An initialized 128-bit floating-point vector of [4 x float] with
+///    all elements set to zero.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_setzero_ps(void)
 {
   return (__m128){ 0, 0, 0, 0 };
 }
 
+/// \brief Stores the upper 64 bits of a 128-bit vector of [4 x float] to a
+///    memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VPEXTRQ / MOVQ instruction.
+///
+/// \param __p
+///    A pointer to a 64-bit memory location.
+/// \param __a
+///    A 128-bit vector of [4 x float] containing the values to be stored.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_storeh_pi(__m64 *__p, __m128 __a)
 {
   __builtin_ia32_storehps((__v2si *)__p, (__v4sf)__a);
 }
 
+/// \brief Stores the lower 64 bits of a 128-bit vector of [4 x float] to a
+///     memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVLPS / MOVLPS instruction.
+///
+/// \param __p
+///    A pointer to a memory location that will receive the float values.
+/// \param __a
+///    A 128-bit vector of [4 x float] containing the values to be stored.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_storel_pi(__m64 *__p, __m128 __a)
 {
   __builtin_ia32_storelps((__v2si *)__p, (__v4sf)__a);
 }
 
+/// \brief Stores the lower 32 bits of a 128-bit vector of [4 x float] to a
+///     memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVSS / MOVSS instruction.
+///
+/// \param __p
+///    A pointer to a 32-bit memory location.
+/// \param __a
+///    A 128-bit vector of [4 x float] containing the value to be stored.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_store_ss(float *__p, __m128 __a)
 {
@@ -1586,6 +1913,18 @@ _mm_store_ss(float *__p, __m128 __a)
   ((struct __mm_store_ss_struct*)__p)->__u = __a[0];
 }
 
+/// \brief Stores float values from a 128-bit vector of [4 x float] to an
+///    unaligned memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVUPS / MOVUPS instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location. The address of the memory
+///    location does not have to be aligned.
+/// \param __a
+///    A 128-bit vector of [4 x float] containing the values to be stored.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_storeu_ps(float *__p, __m128 __a)
 {
@@ -1595,12 +1934,38 @@ _mm_storeu_ps(float *__p, __m128 __a)
   ((struct __storeu_ps*)__p)->__v = __a;
 }
 
+/// \brief Stores the lower 32 bits of a 128-bit vector of [4 x float] into
+///    four contiguous elements in an aligned memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to \c VMOVAPS / MOVAPS + \c shuffling
+///    instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location.
+/// \param __a
+///    A 128-bit vector of [4 x float] whose lower 32 bits are stored to each
+///    of the four contiguous elements pointed by __p.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_store_ps(float *__p, __m128 __a)
 {
   *(__m128*)__p = __a;
 }
 
+/// \brief Stores the lower 32 bits of a 128-bit vector of [4 x float] into
+///    four contiguous elements in an aligned memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to \c VMOVAPS / MOVAPS + \c shuffling
+///    instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location.
+/// \param __a
+///    A 128-bit vector of [4 x float] whose lower 32 bits are stored to each
+///    of the four contiguous elements pointed by __p.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_store1_ps(float *__p, __m128 __a)
 {
@@ -1608,12 +1973,37 @@ _mm_store1_ps(float *__p, __m128 __a)
   _mm_store_ps(__p, __a);
 }
 
+/// \brief Stores float values from a 128-bit vector of [4 x float] to an
+///    aligned memory location.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVAPS / MOVAPS instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location. The address of the memory
+///    location has to be 128-bit aligned.
+/// \param __a
+///    A 128-bit vector of [4 x float] containing the values to be stored.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_store_ps1(float *__p, __m128 __a)
 {
   return _mm_store1_ps(__p, __a);
 }
 
+/// \brief Stores float values from a 128-bit vector of [4 x float] to an
+///    aligned memory location in reverse order.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVAPS / MOVAPS + \c shuffling
+///    instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit memory location. The address of the memory
+///    location has to be 128-bit aligned.
+/// \param __a
+///    A 128-bit vector of [4 x float] containing the values to be stored.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_storer_ps(float *__p, __m128 __a)
 {
@@ -1630,27 +2020,100 @@ _mm_storer_ps(float *__p, __m128 __a)
 /* FIXME: We have to #define this because "sel" must be a constant integer, and
    Sema doesn't do any form of constant propagation yet. */
 
+/// \brief Loads one cache line of data from the specified address to a location
+///    closer to the processor.
+///
+/// \headerfile <x86intrin.h>
+///
+/// \code
+/// void _mm_prefetch(const void * a, const int sel);
+/// \endcode
+///
+/// This intrinsic corresponds to the \c PREFETCHNTA instruction.
+///
+/// \param a
+///    A pointer to a memory location containing a cache line of data.
+/// \param sel
+///    A predefined integer constant specifying the type of prefetch operation:
+///    _MM_HINT_NTA: Move data using the non-temporal access (NTA) hint.
+///    The PREFETCHNTA instruction will be generated.
+///    _MM_HINT_T0: Move data using the T0 hint. The PREFETCHT0 instruction will
+///    be generated.
+///    _MM_HINT_T1: Move data using the T1 hint. The PREFETCHT1 instruction will
+///    be generated.
+///    _MM_HINT_T2: Move data using the T2 hint. The PREFETCHT2 instruction will
+///    be generated.
 #define _mm_prefetch(a, sel) (__builtin_prefetch((void *)(a), 0, (sel)))
 #endif
 
+/// \brief Stores a 64-bit integer in the specified aligned memory location. To
+///    minimize caching, the data is flagged as non-temporal (unlikely to be
+///    used again soon).
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c MOVNTQ instruction.
+///
+/// \param __p
+///    A pointer to an aligned memory location used to store the register value.
+/// \param __a
+///    A 64-bit integer containing the value to be stored.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_stream_pi(__m64 *__p, __m64 __a)
 {
   __builtin_ia32_movntq(__p, __a);
 }
 
+/// \brief Moves packed float values from a 128-bit vector of [4 x float] to a
+///    128-bit aligned memory location. To minimize caching, the data is flagged
+///    as non-temporal (unlikely to be used again soon).
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVNTPS / MOVNTPS instruction.
+///
+/// \param __p
+///    A pointer to a 128-bit aligned memory location that will receive the
+///    integer values.
+/// \param __a
+///    A 128-bit vector of [4 x float] containing the values to be moved.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_stream_ps(float *__p, __m128 __a)
 {
   __builtin_ia32_movntps(__p, (__v4sf)__a);
 }
 
+/// \brief Forces strong memory ordering (serialization) between store
+///    instructions preceding this instruction and store instructions following
+///    this instruction, ensuring the system completes all previous stores
+///    before executing subsequent stores.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c SFENCE instruction.
+///
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_sfence(void)
 {
   __builtin_ia32_sfence();
 }
 
+/// \brief Extracts 16-bit element from a 64-bit vector of [4 x i16] and
+///    returns it, as specified by the immediate integer operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VPEXTRW / PEXTRW instruction.
+///
+/// \param __a
+///    A 64-bit vector of [4 x i16].
+/// \param __n
+///    An immediate integer operand that determines which bits are extracted:
+///    0: Bits [15:0] are copied to the destination.
+///    1: Bits [31:16] are copied to the destination.
+///    2: Bits [47:32] are copied to the destination.
+///    3: Bits [63:48] are copied to the destination.
+/// \returns A 16-bit integer containing the extracted 16 bits of packed data.
 static __inline__ int __DEFAULT_FN_ATTRS
 _mm_extract_pi16(__m64 __a, int __n)
 {
@@ -1658,6 +2121,30 @@ _mm_extract_pi16(__m64 __a, int __n)
   return (unsigned short)__b[__n & 3];
 }
 
+/// \brief Copies data from the 64-bit vector of [4 x i16] to the destination,
+///    and inserts the lower 16-bits of an integer operand at the 16-bit offset
+///    specified by the immediate operand __n.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VPINSRW / PINSRW instruction.
+///
+/// \param __a
+///    A 64-bit vector of [4 x i16].
+/// \param __d
+///    An integer. The lower 16-bit value from this operand is written to the
+///    destination at the offset specified by operand __n.
+/// \param __n
+///    An immediate integer operant that determines which the bits to be used
+///    in the destination.
+///    0: Bits [15:0] are copied to the destination.
+///    1: Bits [31:16] are copied to the destination.
+///    2: Bits [47:32] are copied to the destination.
+///    3: Bits [63:48] are copied to the destination.
+///    The remaining bits in the destination are copied from the corresponding
+///    bits in operand __a.
+/// \returns A 64-bit integer vector containing the copied packed data from the
+///    operands.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_insert_pi16(__m64 __a, int __d, int __n)
 {
@@ -1666,117 +2153,474 @@ _mm_insert_pi16(__m64 __a, int __d, int
    return (__m64)__b;
 }
 
+/// \brief Compares each of the corresponding packed 16-bit integer values of
+///    the 64-bit integer vectors, and writes the greater value to the
+///    corresponding bits in the destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PMAXSW instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector containing the comparison results.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_max_pi16(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_pmaxsw((__v4hi)__a, (__v4hi)__b);
 }
 
+/// \brief Compares each of the corresponding packed 8-bit unsigned integer
+///    values of the 64-bit integer vectors, and writes the greater value to the
+///    corresponding bits in the destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PMAXUB instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector containing the comparison results.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_max_pu8(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_pmaxub((__v8qi)__a, (__v8qi)__b);
 }
 
+/// \brief Compares each of the corresponding packed 16-bit integer values of
+///    the 64-bit integer vectors, and writes the lesser value to the
+///    corresponding bits in the destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PMINSW instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector containing the comparison results.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_min_pi16(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_pminsw((__v4hi)__a, (__v4hi)__b);
 }
 
+/// \brief Compares each of the corresponding packed 8-bit unsigned integer
+///    values of the 64-bit integer vectors, and writes the lesser value to the
+///    corresponding bits in the destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PMINUB instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector containing the comparison results.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_min_pu8(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_pminub((__v8qi)__a, (__v8qi)__b);
 }
 
+/// \brief Takes the most significant bit from each 8-bit element in a 64-bit
+///    integer vector to create a 16-bit mask value. Zero-extends the value to
+///    32-bit integer and writes it to the destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PMOVMSKB instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing the values with bits to be extracted.
+/// \returns The most significant bit from each 8-bit element in the operand,
+///    written to bits [15:0].
 static __inline__ int __DEFAULT_FN_ATTRS
 _mm_movemask_pi8(__m64 __a)
 {
   return __builtin_ia32_pmovmskb((__v8qi)__a);
 }
 
+/// \brief Multiplies packed 16-bit unsigned integer values and writes the
+///    high-order 16 bits of each 32-bit product to the corresponding bits in
+///    the destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PMULHUW instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector containing the products of both operands.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_mulhi_pu16(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_pmulhuw((__v4hi)__a, (__v4hi)__b);
 }
 
+/// \brief Shuffles the 4 16-bit integers from a 64-bit integer vector to the
+///    destination, as specified by the immediate value operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PSHUFW instruction.
+///
+/// \code
+/// __m64 _mm_shuffle_pi16(__m64 a, const int n);
+/// \endcode
+///
+/// \param a
+///    A 64-bit integer vector containing the values to be shuffled.
+/// \param n
+///    An immediate value containing an 8-bit value specifying which elements to
+///    copy from a. The destinations within the 64-bit destination are assigned
+///    values as follows:
+///    Bits [1:0] are used to assign values to bits [15:0] in the destination.
+///    Bits [3:2] are used to assign values to bits [31:16] in the destination.
+///    Bits [5:4] are used to assign values to bits [47:32] in the destination.
+///    Bits [7:6] are used to assign values to bits [63:48] in the destination.
+///    Bit value assignments:
+///    00: assigned from bits [15:0] of a.
+///    01: assigned from bits [31:16] of a.
+///    10: assigned from bits [47:32] of a.
+///    11: assigned from bits [63:48] of a.
+/// \returns A 64-bit integer vector containing the shuffled values.
 #define _mm_shuffle_pi16(a, n) __extension__ ({ \
   (__m64)__builtin_ia32_pshufw((__v4hi)(__m64)(a), (n)); })
 
+/// \brief Conditionally copies the values from each 8-bit element in the first
+///    64-bit integer vector operand to the specified memory location, as
+///    specified by the most significant bit in the corresponding element in the
+///    second 64-bit integer vector operand. To minimize caching, the data is
+///    flagged as non-temporal (unlikely to be used again soon).
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c MASKMOVQ instruction.
+///
+/// \param __d
+///    A 64-bit integer vector containing the values with elements to be copied.
+/// \param __n
+///    A 64-bit integer vector operand. The most significant bit from each 8-bit
+///    element determines whether the corresponding element in operand __d is
+///    copied. If the most significant bit of a given element is 1, the
+///    corresponding element in operand __d is copied.
+/// \param __p
+///    A pointer to a 64-bit memory location that will receive the conditionally
+///    copied integer values. The address of the memory location does not have
+///    to be aligned.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_maskmove_si64(__m64 __d, __m64 __n, char *__p)
 {
   __builtin_ia32_maskmovq((__v8qi)__d, (__v8qi)__n, __p);
 }
 
+/// \brief Computes the rounded averages of the packed unsigned 8-bit integer
+///    values and writes the averages to the corresponding bits in the
+///    destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PAVGB instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector containing the averages of both operands.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_avg_pu8(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_pavgb((__v8qi)__a, (__v8qi)__b);
 }
 
+/// \brief Computes the rounded averages of the packed unsigned 16-bit integer
+///    values and writes the averages to the corresponding bits in the
+///    destination.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PAVGW instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector containing the averages of both operands.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_avg_pu16(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_pavgw((__v4hi)__a, (__v4hi)__b);
 }
 
+/// \brief Subtracts the corresponding 8-bit unsigned integer values of the two
+///    64-bit vector operands and computes the absolute value for each of the
+///    difference. Then sum of the 8 absolute differences is written to the
+///    bits [15:0] of the destination; the remaining bits [63:16] are cleared.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c PSADBW instruction.
+///
+/// \param __a
+///    A 64-bit integer vector containing one of the source operands.
+/// \param __b
+///    A 64-bit integer vector containing one of the source operands.
+/// \returns A 64-bit integer vector whose lower 16 bits contain the sums of the
+///    sets of absolute differences between both operands. The upper bits are
+///    cleared.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_sad_pu8(__m64 __a, __m64 __b)
 {
   return (__m64)__builtin_ia32_psadbw((__v8qi)__a, (__v8qi)__b);
 }
 
+/// \brief Returns the contents of the MXCSR register as a 32-bit unsigned
+///    integer value. There are several groups of macros associated with this
+///    intrinsic, including:
+///    * For checking exception states: _MM_EXCEPT_INVALID, _MM_EXCEPT_DIV_ZERO,
+///      _MM_EXCEPT_DENORM, _MM_EXCEPT_OVERFLOW, _MM_EXCEPT_UNDERFLOW,
+///      _MM_EXCEPT_INEXACT. There is a convenience wrapper
+///      _MM_GET_EXCEPTION_STATE().
+///    * For checking exception masks: _MM_MASK_UNDERFLOW, _MM_MASK_OVERFLOW,
+///      _MM_MASK_INVALID, _MM_MASK_DENORM, _MM_MASK_DIV_ZERO, _MM_MASK_INEXACT.
+///      There is a convenience wrapper _MM_GET_EXCEPTION_MASK().
+///    * For checking rounding modes: _MM_ROUND_NEAREST, _MM_ROUND_DOWN,
+///      _MM_ROUND_UP, _MM_ROUND_TOWARD_ZERO. There is a convenience wrapper
+///      _MM_GET_ROUNDING_MODE(x) where x is one of these macros.
+///    * For checking flush-to-zero mode: _MM_FLUSH_ZERO_ON, _MM_FLUSH_ZERO_OFF.
+///      There is a convenience wrapper _MM_GET_FLUSH_ZERO_MODE().
+///    * For checking denormals-are-zero mode: _MM_DENORMALS_ZERO_ON,
+///      _MM_DENORMALS_ZERO_OFF. There is a convenience wrapper
+///      _MM_GET_DENORMALS_ZERO_MODE().
+///
+///    For example, the expression below checks if an overflow exception has
+///    occurred:
+///      ( _mm_getcsr() & _MM_EXCEPT_OVERFLOW )
+///
+///    The following example gets the current rounding mode:
+///      _MM_GET_ROUNDING_MODE()
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VSTMXCSR / STMXCSR instruction.
+///
+/// \returns A 32-bit unsigned integer containing the contents of the MXCSR
+///    register.
 static __inline__ unsigned int __DEFAULT_FN_ATTRS
 _mm_getcsr(void)
 {
   return __builtin_ia32_stmxcsr();
 }
 
+/// \brief Sets the MXCSR register with the 32-bit unsigned integer value. There
+///    are several groups of macros associated with this intrinsic, including:
+///    * For setting exception states: _MM_EXCEPT_INVALID, _MM_EXCEPT_DIV_ZERO,
+///      _MM_EXCEPT_DENORM, _MM_EXCEPT_OVERFLOW, _MM_EXCEPT_UNDERFLOW,
+///      _MM_EXCEPT_INEXACT. There is a convenience wrapper
+///      _MM_SET_EXCEPTION_STATE(x) where x is one of these macros.
+///    * For setting exception masks: _MM_MASK_UNDERFLOW, _MM_MASK_OVERFLOW,
+///      _MM_MASK_INVALID, _MM_MASK_DENORM, _MM_MASK_DIV_ZERO, _MM_MASK_INEXACT.
+///      There is a convenience wrapper _MM_SET_EXCEPTION_MASK(x) where x is one
+///      of these macros.
+///    * For setting rounding modes: _MM_ROUND_NEAREST, _MM_ROUND_DOWN,
+///      _MM_ROUND_UP, _MM_ROUND_TOWARD_ZERO. There is a convenience wrapper
+///      _MM_SET_ROUNDING_MODE(x) where x is one of these macros.
+///    * For setting flush-to-zero mode: _MM_FLUSH_ZERO_ON, _MM_FLUSH_ZERO_OFF.
+///      There is a convenience wrapper _MM_SET_FLUSH_ZERO_MODE(x) where x is
+///      one of these macros.
+///    * For setting denormals-are-zero mode: _MM_DENORMALS_ZERO_ON,
+///      _MM_DENORMALS_ZERO_OFF. There is a convenience wrapper
+///      _MM_SET_DENORMALS_ZERO_MODE(x) where x is one of these macros.
+///
+///    For example, the following expression causes subsequent floating-point
+///    operations to round up:
+///      _mm_setcsr(_mm_getcsr() | _MM_ROUND_UP)
+///
+///    The following example sets the DAZ and FTZ flags:
+///      void setFlags() {
+///        _MM_SET_FLUSH_ZERO_MODE(_MM_FLUSH_ZERO_ON)
+///        _MM_SET_DENORMALS_ZERO_MODE(_MM_DENORMALS_ZERO_ON)
+///      }
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VLDMXCSR / LDMXCSR instruction.
+///
+/// \param __i
+///    A 32-bit unsigned integer value to be written to the MXCSR register.
 static __inline__ void __DEFAULT_FN_ATTRS
 _mm_setcsr(unsigned int __i)
 {
   __builtin_ia32_ldmxcsr(__i);
 }
 
+/// \brief Selects 4 float values from the 128-bit operands of [4 x float], as
+///    specified by the immediate value operand.
+///
+/// \headerfile <x86intrin.h>
+///
+/// \code
+/// __m128 _mm_shuffle_ps(__m128 a, __m128 b, const int mask);
+/// \endcode
+///
+/// This intrinsic corresponds to the \c VSHUFPS / SHUFPS instruction.
+///
+/// \param a
+///    A 128-bit vector of [4 x float].
+/// \param b
+///    A 128-bit vector of [4 x float].
+/// \param mask
+///    An immediate value containing an 8-bit value specifying which elements to
+///    copy from a and b.
+///    Bits [3:0] specify the values copied from operand a.
+///    Bits [7:4] specify the values copied from operand b. The destinations
+///    within the 128-bit destination are assigned values as follows:
+///    Bits [1:0] are used to assign values to bits [31:0] in the destination.
+///    Bits [3:2] are used to assign values to bits [63:32] in the destination.
+///    Bits [5:4] are used to assign values to bits [95:64] in the destination.
+///    Bits [7:6] are used to assign values to bits [127:96] in the destination.
+///    Bit value assignments:
+///    00: Bits [31:0] copied from the specified operand.
+///    01: Bits [63:32] copied from the specified operand.
+///    10: Bits [95:64] copied from the specified operand.
+///    11: Bits [127:96] copied from the specified operand.
+/// \returns A 128-bit vector of [4 x float] containing the shuffled values.
 #define _mm_shuffle_ps(a, b, mask) __extension__ ({ \
   (__m128)__builtin_shufflevector((__v4sf)(__m128)(a), (__v4sf)(__m128)(b), \
                                   (mask) & 0x3, ((mask) & 0xc) >> 2, \
                                   (((mask) & 0x30) >> 4) + 4, \
                                   (((mask) & 0xc0) >> 6) + 4); })
 
+/// \brief Unpacks the high-order (index 2,3) values from two 128-bit vectors of
+///    [4 x float] and interleaves them into a 128-bit vector of [4 x
+///    float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VUNPCKHPS / UNPCKHPS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+///    Bits [95:64] are written to bits [31:0] of the destination.
+///    Bits [127:96] are written to bits [95:64] of the destination.
+/// \param __b
+///    A 128-bit vector of [4 x float].
+///    Bits [95:64] are written to bits [63:32] of the destination.
+///    Bits [127:96] are written to bits [127:96] of the destination.
+/// \returns A 128-bit vector of [4 x float] containing the interleaved values.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_unpackhi_ps(__m128 __a, __m128 __b)
 {
   return __builtin_shufflevector((__v4sf)__a, (__v4sf)__b, 2, 6, 3, 7);
 }
 
+/// \brief Unpacks the low-order (index 0,1) values from two 128-bit vectors of
+///    [4 x float] and interleaves them into a 128-bit vector of [4 x
+///    float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VUNPCKLPS / UNPCKLPS instruction.
+///
+/// \param __a
+///    A 128-bit vector of [4 x float].
+///    Bits [31:0] are written to bits [31:0] of the destination.
+///    Bits [63:32] are written to bits [95:64] of the destination.
+/// \param __b
+///    A 128-bit vector of [4 x float].
+///    Bits [31:0] are written to bits [63:32] of the destination.
+///    Bits [63:32] are written to bits [127:96] of the destination.
+/// \returns A 128-bit vector of [4 x float] containing the interleaved values.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_unpacklo_ps(__m128 __a, __m128 __b)
 {
   return __builtin_shufflevector((__v4sf)__a, (__v4sf)__b, 0, 4, 1, 5);
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float]. The lower
+///    32 bits are set to the lower 32 bits of the second parameter. The upper
+///    96 bits are set to the upper 96 bits of the first parameter.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVSS / MOVSS instruction.
+///
+/// \param __a
+///    A 128-bit floating-point vector of [4 x float]. The upper 96 bits are
+///    written to the upper 96 bits of the result.
+/// \param __b
+///    A 128-bit floating-point vector of [4 x float]. The lower 32 bits are
+///    written to the lower 32 bits of the result.
+/// \returns A 128-bit floating-point vector of [4 x float].
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_move_ss(__m128 __a, __m128 __b)
 {
   return __builtin_shufflevector((__v4sf)__a, (__v4sf)__b, 4, 1, 2, 3);
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float]. The lower
+///    64 bits are set to the upper 64 bits of the second parameter. The upper
+///    64 bits are set to the upper 64 bits of the first parameter.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VUNPCKHPD / UNPCKHPD instruction.
+///
+/// \param __a
+///    A 128-bit floating-point vector of [4 x float]. The upper 64 bits are
+///    written to the upper 64 bits of the result.
+/// \param __b
+///    A 128-bit floating-point vector of [4 x float]. The upper 64 bits are
+///    written to the lower 64 bits of the result.
+/// \returns A 128-bit floating-point vector of [4 x float].
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_movehl_ps(__m128 __a, __m128 __b)
 {
   return __builtin_shufflevector((__v4sf)__a, (__v4sf)__b, 6, 7, 2, 3);
 }
 
+/// \brief Constructs a 128-bit floating-point vector of [4 x float]. The lower
+///    64 bits are set to the lower 64 bits of the first parameter. The upper
+///    64 bits are set to the lower 64 bits of the second parameter.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VUNPCKLPD / UNPCKLPD instruction.
+///
+/// \param __a
+///    A 128-bit floating-point vector of [4 x float]. The lower 64 bits are
+///    written to the lower 64 bits of the result.
+/// \param __b
+///    A 128-bit floating-point vector of [4 x float]. The lower 64 bits are
+///    written to the upper 64 bits of the result.
+/// \returns A 128-bit floating-point vector of [4 x float].
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_movelh_ps(__m128 __a, __m128 __b)
 {
   return __builtin_shufflevector((__v4sf)__a, (__v4sf)__b, 0, 1, 4, 5);
 }
 
+/// \brief Converts a 64-bit vector of [4 x i16] into a 128-bit vector of [4 x
+///    float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPI2PS + \c COMPOSITE instruction.
+///
+/// \param __a
+///    A 64-bit vector of [4 x i16]. The elements of the destination are copied
+///    from the corresponding elements in this operand.
+/// \returns A 128-bit vector of [4 x float] containing the copied and converted
+///    values from the operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtpi16_ps(__m64 __a)
 {
@@ -1795,6 +2639,18 @@ _mm_cvtpi16_ps(__m64 __a)
   return __r;
 }
 
+/// \brief Converts a 64-bit vector of 16-bit unsigned integer values into a
+///    128-bit vector of [4 x float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPI2PS + \c COMPOSITE instruction.
+///
+/// \param __a
+///    A 64-bit vector of 16-bit unsigned integer values. The elements of the
+///    destination are copied from the corresponding elements in this operand.
+/// \returns A 128-bit vector of [4 x float] containing the copied and converted
+///    values from the operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtpu16_ps(__m64 __a)
 {
@@ -1812,6 +2668,18 @@ _mm_cvtpu16_ps(__m64 __a)
   return __r;
 }
 
+/// \brief Converts the lower four 8-bit values from a 64-bit vector of [8 x i8]
+///    into a 128-bit vector of [4 x float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPI2PS + \c COMPOSITE instruction.
+///
+/// \param __a
+///    A 64-bit vector of [8 x i8]. The elements of the destination are copied
+///    from the corresponding lower 4 elements in this operand.
+/// \returns A 128-bit vector of [4 x float] containing the copied and converted
+///    values from the operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtpi8_ps(__m64 __a)
 {
@@ -1824,6 +2692,19 @@ _mm_cvtpi8_ps(__m64 __a)
   return _mm_cvtpi16_ps(__b);
 }
 
+/// \brief Converts the lower four unsigned 8-bit integer values from a 64-bit
+///    vector of [8 x u8] into a 128-bit vector of [4 x float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPI2PS + \c COMPOSITE instruction.
+///
+/// \param __a
+///    A 64-bit vector of unsigned 8-bit integer values. The elements of the
+///    destination are copied from the corresponding lower 4 elements in this
+///    operand.
+/// \returns A 128-bit vector of [4 x float] containing the copied and converted
+///    values from the source operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtpu8_ps(__m64 __a)
 {
@@ -1835,6 +2716,22 @@ _mm_cvtpu8_ps(__m64 __a)
   return _mm_cvtpi16_ps(__b);
 }
 
+/// \brief Converts the two 32-bit signed integer values from each 64-bit vector
+///    operand of [2 x i32] into a 128-bit vector of [4 x float].
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPI2PS + \c COMPOSITE instruction.
+///
+/// \param __a
+///    A 64-bit vector of [2 x i32]. The lower elements of the destination are
+///    copied from the elements in this operand.
+/// \param __b
+///    A 64-bit vector of [2 x i32]. The upper elements of the destination are
+///    copied from the elements in this operand.
+/// \returns A 128-bit vector of [4 x float] whose lower 64 bits contain the
+///    copied and converted values from the first operand. The upper 64 bits
+///    contain the copied and converted values from the second operand.
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_cvtpi32x2_ps(__m64 __a, __m64 __b)
 {
@@ -1847,6 +2744,22 @@ _mm_cvtpi32x2_ps(__m64 __a, __m64 __b)
   return _mm_cvtpi32_ps(__c, __a);
 }
 
+/// \brief Converts each single-precision floating-point element of a 128-bit
+///    floating-point vector of [4 x float] into a 16-bit signed integer, and
+///    packs the results into a 64-bit integer vector of [4 x i16]. If the
+///    floating-point element is NaN or infinity, or if the floating-point
+///    element is greater than 0x7FFFFFFF or less than -0x8000, it is converted
+///    to 0x8000. Otherwise if the floating-point element is greater
+///    than 0x7FFF, it is converted to 0x7FFF.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPS2PI + \c COMPOSITE instruction.
+///
+/// \param __a
+///    A 128-bit floating-point vector of [4 x float].
+/// \returns A 64-bit integer vector of [4 x i16] containing the converted
+///    values.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_cvtps_pi16(__m128 __a)
 {
@@ -1859,6 +2772,23 @@ _mm_cvtps_pi16(__m128 __a)
   return _mm_packs_pi32(__b, __c);
 }
 
+/// \brief Converts each single-precision floating-point element of a 128-bit
+///    floating-point vector of [4 x float] into an 8-bit signed integer, and
+///    packs the results into the lower 32 bits of a 64-bit integer vector of
+///    [8 x i8]. The upper 32 bits of the vector are set to 0. If the
+///    floating-point element is NaN or infinity, or if the floating-point
+///    element is greater than 0x7FFFFFFF or less than -0x80, it is converted
+///    to 0x80. Otherwise if the floating-point element is greater
+///    than 0x7F, it is converted to 0x7F.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c CVTPS2PI + \c COMPOSITE instruction.
+///
+/// \param __a
+///    128-bit floating-point vector of [4 x float].
+/// \returns A 64-bit integer vector of [8 x i8]. The lower 32 bits contain the
+///    converted values and the uppper 32 bits are set to zero.
 static __inline__ __m64 __DEFAULT_FN_ATTRS
 _mm_cvtps_pi8(__m128 __a)
 {
@@ -1870,6 +2800,20 @@ _mm_cvtps_pi8(__m128 __a)
   return _mm_packs_pi16(__b, __c);
 }
 
+/// \brief Extracts the sign bits from each single-precision floating-point
+///    element of a 128-bit floating-point vector of [4 x float] and returns the
+///    sign bits in bits [0:3] of the result. Bits [31:4] of the result are set
+///    to zero.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsic corresponds to the \c VMOVMSKPS / MOVMSKPS instruction.
+///
+/// \param __a
+///    A 128-bit floating-point vector of [4 x float].
+/// \returns A 32-bit integer value. Bits [3:0] contain the sign bits from each
+///    single-precision floating-point element of the parameter. Bits [31:4] are
+///    set to zero.
 static __inline__ int __DEFAULT_FN_ATTRS
 _mm_movemask_ps(__m128 __a)
 {




More information about the cfe-commits mailing list