src/gallium/auxiliary/util/u_sse.h

   1 /**************************************************************************
   2  *
   3  * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
   4  * All Rights Reserved.
   5  *
   6  * Permission is hereby granted, free of charge, to any person obtaining a
   7  * copy of this software and associated documentation files (the
   8  * "Software"), to deal in the Software without restriction, including
   9  * without limitation the rights to use, copy, modify, merge, publish,
  10  * distribute, sub license, and/or sell copies of the Software, and to
  11  * permit persons to whom the Software is furnished to do so, subject to
  12  * the following conditions:
  13  *
  14  * The above copyright notice and this permission notice (including the
  15  * next paragraph) shall be included in all copies or substantial portions
  16  * of the Software.
  17  *
  18  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  19  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  20  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
  21  * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
  22  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  23  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  24  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  25  *
  26  **************************************************************************/
  27
  28 /**
  29  * @file
  30  * SSE intrinsics portability header.
  31  *
  32  * Although the SSE intrinsics are support by all modern x86 and x86-64
  33  * compilers, there are some intrisincs missing in some implementations
  34  * (especially older MSVC versions). This header abstracts that away.
  35  */
  36
  37 #ifndef U_SSE_H_
  38 #define U_SSE_H_
  39
  40 #include "pipe/p_config.h"
  41
  42 #if defined(PIPE_ARCH_SSE)
  43
  44 #include <emmintrin.h>
  45
  46
  47 /* MSVC before VC8 does not support the _mm_castxxx_yyy */
  48 #if defined(_MSC_VER) && _MSC_VER < 1500
  49
  50 union __declspec(align(16)) m128_types {
  51    __m128 m128;
  52    __m128i m128i;
  53    __m128d m128d;
  54 };
  55
  56 static __inline __m128
  57 _mm_castsi128_ps(__m128i a)
  58 {
  59    union m128_types u;
  60    u.m128i = a;
  61    return u.m128;
  62 }
  63
  64 static __inline __m128i
  65 _mm_castps_si128(__m128 a)
  66 {
  67    union m128_types u;
  68    u.m128 = a;
  69    return u.m128i;
  70 }
  71
  72 #endif /* defined(_MSC_VER) && _MSC_VER < 1500 */
  73
  74
  75 #if defined(PIPE_ARCH_SSSE3)
  76
  77 #include <tmmintrin.h>
  78
  79 #else /* !PIPE_ARCH_SSSE3 */
  80
  81 #include <emmintrin.h>
  82
  83 /**
  84  * Describe _mm_shuffle_epi8() with gcc extended inline assembly, for cases
  85  * where -mssse3 is not supported/enabled.
  86  *
  87  * MSVC will never get in here as its intrinsics support do not rely on
  88  * compiler command line options.
  89  */
  90 static __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __artificial__))
  91 _mm_shuffle_epi8(__m128i a, __m128i mask)
  92 {
  93     __m128i result;
  94     __asm__("pshufb %1, %0"
  95             : "=x" (result)
  96             : "xm" (mask), "0" (a));
  97     return result;
  98 }
  99
 100 #endif /* !PIPE_ARCH_SSSE3 */
 101
 102
 103 #endif /* PIPE_ARCH_X86 || PIPE_ARCH_X86_64 */
 104
 105 #endif /* U_SSE_H_ */