libclc/generic/lib/math/ilogb.cl

   1 /*
   2  * Copyright (c) 2015 Advanced Micro Devices, Inc.
   3  * Copyright (c) 2016 Aaron Watry
   4  *
   5  * Permission is hereby granted, free of charge, to any person obtaining a copy
   6  * of this software and associated documentation files (the "Software"), to deal
   7  * in the Software without restriction, including without limitation the rights
   8  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
   9  * copies of the Software, and to permit persons to whom the Software is
  10  * furnished to do so, subject to the following conditions:
  11  *
  12  * The above copyright notice and this permission notice shall be included in
  13  * all copies or substantial portions of the Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  18  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  20  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  21  * THE SOFTWARE.
  22  */
  23
  24 #include "math.h"
  25 #include <clc/clc.h>
  26 #include <clc/clcmacro.h>
  27
  28 _CLC_OVERLOAD _CLC_DEF int ilogb(float x) {
  29     uint ux = as_uint(x);
  30     uint ax = ux & EXSIGNBIT_SP32;
  31     int rs = -118 - (int) clz(ux & MANTBITS_SP32);
  32     int r = (int) (ax >> EXPSHIFTBITS_SP32) - EXPBIAS_SP32;
  33     r = ax < 0x00800000U ? rs : r;
  34     r = ax == 0 ? FP_ILOGB0 : r;
  35
  36     // We could merge those 2 tests and have:
  37     //
  38     //    r = ax >= EXPBITS_SP32 ? 0x7fffffff : r
  39     //
  40     // since FP_ILOGBNAN is set to INT_MAX, but it's clearer this way and
  41     // FP_ILOGBNAN can change without requiring changes to ilogb() code.
  42     r = ax > EXPBITS_SP32 ? FP_ILOGBNAN : r;
  43     r = ax == EXPBITS_SP32 ? 0x7fffffff : r;
  44     return r;
  45 }
  46
  47 _CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, int, ilogb, float);
  48
  49 #ifdef cl_khr_fp64
  50 #pragma OPENCL EXTENSION cl_khr_fp64 : enable
  51
  52 _CLC_OVERLOAD _CLC_DEF int ilogb(double x) {
  53     ulong ux = as_ulong(x);
  54     ulong ax = ux & ~SIGNBIT_DP64;
  55     int r = (int) (ax >> EXPSHIFTBITS_DP64) - EXPBIAS_DP64;
  56     int rs = -1011 - (int) clz(ax & MANTBITS_DP64);
  57     r = ax < 0x0010000000000000UL ? rs : r;
  58     r = ax == 0UL ? FP_ILOGB0 : r;
  59
  60     // We could merge those 2 tests and have:
  61     //
  62     //    r = ax >= 0x7ff0000000000000UL ? 0x7fffffff : r
  63     //
  64     // since FP_ILOGBNAN is set to INT_MAX, but it's clearer this way and
  65     // FP_ILOGBNAN can change without requiring changes to ilogb() code.
  66     r = ax > 0x7ff0000000000000UL ? FP_ILOGBNAN : r;
  67     r = ax == 0x7ff0000000000000UL ? 0x7fffffff : r;
  68     return r;
  69 }
  70
  71 _CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, int, ilogb, double);
  72
  73 #endif // cl_khr_fp64
  74
  75 #ifdef cl_khr_fp16
  76
  77 #pragma OPENCL EXTENSION cl_khr_fp16 : enable
  78
  79 _CLC_OVERLOAD _CLC_DEF int ilogb(half x) {
  80     return ilogb((float)x);
  81 }
  82
  83 _CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, int, ilogb, half);
  84
  85 #endif