libclc/generic/lib/math/clc_tanpi.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 #include <clc/clc.h>
  23 #include <clc/clcmacro.h>
  24
  25 #include "math.h"
  26 #include "sincos_helpers.h"
  27 #include "tables.h"
  28
  29 _CLC_DEF _CLC_OVERLOAD float __clc_tanpi(float x)
  30 {
  31     int ix = as_int(x);
  32     int xsgn = ix & 0x80000000;
  33     int xnsgn = xsgn ^ 0x80000000;
  34     ix ^= xsgn;
  35     float ax = as_float(ix);
  36     int iax = (int)ax;
  37     float r = ax - iax;
  38     int xodd = xsgn ^ (iax & 0x1 ? 0x80000000 : 0);
  39
  40     // Initialize with return for +-Inf and NaN
  41     int ir = 0x7fc00000;
  42
  43     // 2^24 <= |x| < Inf, the result is always even integer
  44     ir = ix < 0x7f800000 ? xsgn : ir;
  45
  46     // 2^23 <= |x| < 2^24, the result is always integer
  47     ir = ix < 0x4b800000 ? xodd : ir;
  48
  49     // 0x1.0p-7 <= |x| < 2^23, result depends on which 0.25 interval
  50
  51     // r < 1.0
  52     float a = 1.0f - r;
  53     int e = 0;
  54     int s = xnsgn;
  55
  56     // r <= 0.75
  57     int c = r <= 0.75f;
  58     a = c ? r - 0.5f : a;
  59     e = c ? 1 : e;
  60     s = c ? xsgn : s;
  61
  62     // r < 0.5
  63     c = r < 0.5f;
  64     a = c ? 0.5f - r : a;
  65     s = c ? xnsgn : s;
  66
  67     // 0 < r <= 0.25
  68     c = r <= 0.25f;
  69     a = c ? r : a;
  70     e = c ? 0 : e;
  71     s = c ? xsgn : s;
  72
  73     float t = __clc_tanf_piby4(a * M_PI_F, 0);
  74     float tr = -native_recip(t);
  75     int jr = s ^ as_int(e ? tr : t);
  76
  77     jr = r == 0.5f ? xodd | 0x7f800000 : jr;
  78
  79     ir = ix < 0x4b000000 ? jr : ir;
  80
  81     return as_float(ir);
  82 }
  83 _CLC_UNARY_VECTORIZE(_CLC_DEF _CLC_OVERLOAD, float, __clc_tanpi, float);
  84
  85 #ifdef cl_khr_fp64
  86 #include "sincosD_piby4.h"
  87
  88 _CLC_DEF _CLC_OVERLOAD double __clc_tanpi(double x)
  89 {
  90     long ix = as_long(x);
  91     long xsgn = ix & 0x8000000000000000L;
  92     long xnsgn = xsgn ^ 0x8000000000000000L;
  93     ix ^= xsgn;
  94     double ax = as_double(ix);
  95     long iax = (long)ax;
  96     double r = ax - iax;
  97     long xodd = xsgn ^ (iax & 0x1 ? 0x8000000000000000L : 0L);
  98
  99     // Initialize with return for +-Inf and NaN
 100     long ir = 0x7ff8000000000000L;
 101
 102     // 2^53 <= |x| < Inf, the result is always even integer
 103     ir = ix < 0x7ff0000000000000L ? xsgn : ir;
 104
 105     // 2^52 <= |x| < 2^53, the result is always integer
 106     ir = ix < 0x4340000000000000L ? xodd : ir;
 107
 108     // 0x1.0p-14 <= |x| < 2^53, result depends on which 0.25 interval
 109
 110     // r < 1.0
 111     double a = 1.0 - r;
 112     int e = 0;
 113     long s = xnsgn;
 114
 115     // r <= 0.75
 116     int c = r <= 0.75;
 117     double t = r - 0.5;
 118     a = c ? t : a;
 119     e = c ? 1 : e;
 120     s = c ? xsgn : s;
 121
 122     // r < 0.5
 123     c = r < 0.5;
 124     t = 0.5 - r;
 125     a = c ? t : a;
 126     s = c ? xnsgn : s;
 127
 128     // r <= 0.25
 129     c = r <= 0.25;
 130     a = c ? r : a;
 131     e = c ? 0 : e;
 132     s = c ? xsgn : s;
 133
 134     double api = a * M_PI;
 135     double2 tt = __clc_tan_piby4(api, 0.0);
 136     long jr = s ^ as_long(e ? tt.hi : tt.lo);
 137
 138     long si = xodd | 0x7ff0000000000000L;
 139     jr = r == 0.5 ? si : jr;
 140
 141     ir = ix < 0x4330000000000000L ? jr : ir;
 142
 143     return as_double(ir);
 144 }
 145 _CLC_UNARY_VECTORIZE(_CLC_DEF _CLC_OVERLOAD, double, __clc_tanpi, double);
 146 #endif