libclc/generic/lib/math/sinpi.cl

   1 /*
   2  * Copyright (c) 2014 Advanced Micro Devices, Inc.
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a copy
   5  * of this software and associated documentation files (the "Software"), to deal
   6  * in the Software without restriction, including without limitation the rights
   7  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
   8  * copies of the Software, and to permit persons to whom the Software is
   9  * furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice shall be included in
  12  * all copies or substantial portions of the Software.
  13  *
  14  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  15  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  16  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  17  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  18  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  19  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  20  * THE SOFTWARE.
  21  */
  22
  23 #include <clc/clc.h>
  24 #include <clc/clcmacro.h>
  25
  26 #include "math.h"
  27 #include "sincospiF_piby4.h"
  28 #ifdef cl_khr_fp64
  29 #include "sincosD_piby4.h"
  30 #endif
  31
  32 _CLC_OVERLOAD _CLC_DEF float sinpi(float x)
  33 {
  34     int ix = as_int(x);
  35     int xsgn = ix & 0x80000000;
  36     ix ^= xsgn;
  37     float ax = as_float(ix);
  38     int iax = (int)ax;
  39     float r = ax - iax;
  40     int xodd = xsgn ^ (iax & 0x1 ? 0x80000000 : 0);
  41
  42     // Initialize with return for +-Inf and NaN
  43     int ir = 0x7fc00000;
  44
  45     // 2^23 <= |x| < Inf, the result is always integer
  46     ir = ix < 0x7f800000 ? xsgn : ir;
  47
  48     // 0x1.0p-7 <= |x| < 2^23, result depends on which 0.25 interval
  49
  50     // r < 1.0
  51     float a = 1.0f - r;
  52     int e = 0;
  53
  54     // r <= 0.75
  55     int c = r <= 0.75f;
  56     a = c ? r - 0.5f : a;
  57     e = c ? 1 : e;
  58
  59     // r < 0.5
  60     c = r < 0.5f;
  61     a = c ? 0.5f - r : a;
  62
  63     // 0 < r <= 0.25
  64     c = r <= 0.25f;
  65     a = c ? r : a;
  66     e = c ? 0 : e;
  67
  68     float2 t = __libclc__sincosf_piby4(a * M_PI_F);
  69     int jr = xodd ^ as_int(e ? t.hi : t.lo);
  70
  71     ir = ix < 0x4b000000 ? jr : ir;
  72
  73     return as_float(ir);
  74 }
  75
  76 _CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, float, sinpi, float);
  77
  78 #ifdef cl_khr_fp64
  79
  80 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
  81
  82 _CLC_OVERLOAD _CLC_DEF double sinpi(double x)
  83 {
  84     long ix = as_long(x);
  85     long xsgn = ix & 0x8000000000000000L;
  86     ix ^= xsgn;
  87     double ax = as_double(ix);
  88     long iax = (long)ax;
  89     double r = ax - (double)iax;
  90     long xodd = xsgn ^ (iax & 0x1L ? 0x8000000000000000L : 0L);
  91
  92     // Initialize with return for +-Inf and NaN
  93     long ir = 0x7ff8000000000000L;
  94
  95     // 2^23 <= |x| < Inf, the result is always integer
  96     ir = ix < 0x7ff0000000000000 ? xsgn : ir;
  97
  98     // 0x1.0p-7 <= |x| < 2^23, result depends on which 0.25 interval
  99
 100     // r < 1.0
 101     double a = 1.0 - r;
 102     int e = 0;
 103
 104     //  r <= 0.75
 105     int c = r <= 0.75;
 106     double t = r - 0.5;
 107     a = c ? t : a;
 108     e = c ? 1 : e;
 109
 110     // r < 0.5
 111     c = r < 0.5;
 112     t = 0.5 - r;
 113     a = c ? t : a;
 114
 115     // r <= 0.25
 116     c = r <= 0.25;
 117     a = c ? r : a;
 118     e = c ? 0 : e;
 119
 120     double api = a * M_PI;
 121     double2 sc = __libclc__sincos_piby4(api, 0.0);
 122     long jr = xodd ^ as_long(e ? sc.hi : sc.lo);
 123
 124     ir = ax < 0x1.0p+52 ? jr : ir;
 125
 126     return as_double(ir);
 127 }
 128
 129 _CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, double, sinpi, double)
 130
 131 #endif
 132
 133 _CLC_DEFINE_UNARY_BUILTIN_FP16(sinpi)