src/gallium/auxiliary/gallivm/soabuiltins.c

   1 /**************************************************************************
   2  *
   3  * Copyright 2007 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   * This file is compiled with clang into the LLVM bitcode
  30   *
  31   * Authors:
  32   *   Zack Rusin zack@tungstengraphics.com
  33   */
  34 typedef __attribute__(( ocu_vector_type(4) )) float float4;
  35
  36 void dp3(float4 *res,
  37          float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w,
  38          float4 tmp1x, float4 tmp1y, float4 tmp1z, float4 tmp1w)
  39 {
  40    float4 dot = (tmp0x * tmp1x) + (tmp0y * tmp1y) +
  41                 (tmp0z * tmp1z);
  42
  43    res[0] = dot;
  44    res[1] = dot;
  45    res[2] = dot;
  46    res[3] = dot;
  47 }
  48
  49
  50 void dp4(float4 *res,
  51          float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w,
  52          float4 tmp1x, float4 tmp1y, float4 tmp1z, float4 tmp1w)
  53 {
  54    float4 dot = (tmp0x * tmp1x) + (tmp0y * tmp1y) +
  55                 (tmp0z * tmp1z) + (tmp0w * tmp1w);
  56
  57    res[0] = dot;
  58    res[1] = dot;
  59    res[2] = dot;
  60    res[3] = dot;
  61 }
  62
  63 extern float powf(float num, float p);
  64
  65 void pow(float4 *res,
  66          float4 tmp0x, float4 tmp0y, float4 tmp0z, float4 tmp0w,
  67          float4 tmp1x, float4 tmp1y, float4 tmp1z, float4 tmp1w)
  68 {
  69    float4 p;
  70    p.x = powf(tmp0x.x, tmp1x.x);
  71    p.y = powf(tmp0x.y, tmp1x.y);
  72    p.z = powf(tmp0x.z, tmp1x.z);
  73    p.w = powf(tmp0x.w, tmp1x.w);
  74
  75    res[0] = p;
  76    res[1] = p;
  77    res[2] = p;
  78    res[3] = p;
  79 }
  80
  81 #if 0
  82 void yo(float4 *out, float4 *in)
  83 {
  84    float4 res[4];
  85
  86    dp3(res, in[0], in[1], in[2], in[3],
  87        in[4], in[5], in[6], in[7]);
  88    out[1] = res[1];
  89 }
  90 #endif